2012 STATE OF THE MARKET REPORT
FOR THE
ERCOT WHOLESALE ELECTRICITY MARKETS
POTOMAC ECONOMICS, LTD.
Independent Market Monitor for the
ERCOT Wholesale Market
June 2013
ERCOT 2012 State of the Market Report
TABLE OF CONTENTS
Executive Summary ....................................................................................................................... i
A.
B.
C.
D.
E.
F.
G.
H.
I.
Review of Real-Time Market Outcomes ........................................................................... 1
A.
B.
C.
D.
E.
II.
Introduction................................................................................................................... i
Review of Real-Time Market Outcomes ..................................................................... ii
Review of Day-Ahead Market Outcomes................................................................... vi
Transmission and Congestion ..................................................................................... ix
Load and Generation................................................................................................. xiii
Resource Adequacy ................................................................................................ xviii
Analysis of Competitive Performance ......................................................................xxv
Recommendations................................................................................................... xxix
Real-Time Market Prices ..............................................................................................1
Real-Time Prices Adjusted for Fuel Price Changes .....................................................9
Real-Time Price Volatility ..........................................................................................13
Prices at the System-Wide Offer Cap .........................................................................15
Mitigation ...................................................................................................................18
Review of Day-Ahead Market Outcomes ........................................................................ 21
A.
B.
C.
D.
Day-Ahead Market Prices...........................................................................................21
Day-Ahead Market Volumes ......................................................................................24
Point to Point Obligations ...........................................................................................26
Ancillary Services Market ..........................................................................................29
III. Transmission and Congestion .......................................................................................... 37
A.
B.
C.
D.
IV.
Load and Generation ........................................................................................................ 55
A.
B.
V.
ERCOT Loads in 2012 ...............................................................................................55
Generation Capacity in ERCOT .................................................................................58
Resource Adequacy ........................................................................................................... 73
A.
B.
C.
VI.
Summary of Congestion .............................................................................................37
Real-Time Constraints ................................................................................................39
Day-Ahead Constraints ...............................................................................................45
Congestion Rights Market ..........................................................................................47
Net Revenue Analysis.................................................................................................73
Effectiveness of the Scarcity Pricing Mechanism ......................................................79
Demand Response Capability .....................................................................................91
Analysis of Competitive Performance ............................................................................. 95
A.
B.
Structural Market Power Indicators ............................................................................95
Evaluation of Supplier Conduct................................................................................102
ERCOT 2012 State of the Market Report
LIST OF FIGURES
Figure 1: Average All-in Price for Electricity in ERCOT ............................................................. 1
Figure 2: Average Real-Time Energy Market Prices .................................................................... 3
Figure 3: Comparison of All-in Prices across Markets .................................................................. 4
Figure 4: ERCOT Price Duration Curve ........................................................................................ 5
Figure 5: ERCOT Price Duration Curve ........................................................................................ 5
Figure 6: Average Real-Time Energy Prices and Number of Price Spikes ................................... 6
Figure 7: Zonal Price Duration Curves .......................................................................................... 8
Figure 8: West Zone and ERCOT Price Duration Curves ............................................................. 8
Figure 9: Implied Marginal Heat Rate Duration Curve – All hours .............................................. 9
Figure 10: Implied Marginal Heat Rate Duration Curve – Top five percent of hours .............. 10
Figure 11: Monthly Average Implied Heat Rates ........................................................................ 11
Figure 12: Heat Rate and Load Relationship ............................................................................... 12
Figure 13: Typical Energy Offers ................................................................................................. 13
Figure 14: Real-Time Energy Price Volatility (May – August) .................................................. 14
Figure 15: Duration of Prices at the System-Wide Offer Cap ..................................................... 15
Figure 16: Load, Reserves and Prices: August 2011 and August 2012 ....................................... 16
Figure 17: Mitigated Capacity by Load Level ............................................................................. 18
Figure 18: Capacity Subject to Mitigation ................................................................................... 19
Figure 19: Convergence between Forward and Real-Time Energy Prices .................................. 23
Figure 20: Day-Ahead and Real-Time Prices by Zone ................................................................ 24
Figure 21: Volume of Day-Ahead Market Activity by Month .................................................... 25
Figure 22: Volume of Day-Ahead Market Activity by Hour ...................................................... 26
Figure 23: Point to Point Obligation Volume .............................................................................. 27
Figure 24: Average Profitability of Point to Point Obligations ................................................... 28
Figure 25: Point to Point Obligation Charges and Payments ...................................................... 29
Figure 26: Ancillary Service Capacity......................................................................................... 30
Figure 27: Ancillary Service Prices ............................................................................................. 31
Figure 28: Ancillary Service Costs per MWh of Load ................................................................ 32
Figure 29: Frequency of SASM Clearing .................................................................................... 33
Figure 30: Ancillary Service Quantities Procured in SASM ....................................................... 35
Figure 31: Frequency of Active Constraints ................................................................................ 38
Figure 32: Top Ten Real-Time Constraints ................................................................................. 39
Figure 33: Real-Time Congestion Costs ...................................................................................... 41
Figure 34: Most Frequent Real-Time Constraints ....................................................................... 43
Figure 35: Utilization of the West to North Interface Constraint ................................................ 44
Figure 36: Top Ten Day-Ahead Constraints ................................................................................ 45
Figure 37: Most Frequent Day-Ahead Constraints ...................................................................... 46
Figure 38: Day-Ahead Congestion Costs .................................................................................... 47
Figure 39: CRR Auction Revenue ............................................................................................... 48
Figure 40: CRR Auction Revenue and Payment Received ......................................................... 49
Figure 41: CRR Auction Revenue, Payments and Congestion Rent ........................................... 50
Figure 42: Day-Ahead and Real-Time Congestion Payments and Rent...................................... 51
Figure 43: Real-Time Congestion Payments ............................................................................... 52
Figure 44: West Hub to North Load Zone Price Spreads ............................................................ 53
ERCOT 2012 State of the Market Report
Figure 45:
Figure 46:
Figure 47:
Figure 48:
Figure 49:
Figure 50:
Figure 51:
Figure 52:
Figure 53:
Figure 54:
Figure 55:
Figure 56:
Figure 57:
Figure 58:
Figure 59:
Figure 60:
Figure 61:
Figure 62:
Figure 63:
Figure 64:
Figure 65:
Figure 66:
Figure 67:
Figure 68:
Figure 69:
Figure 70:
Figure 71:
Figure 72:
Figure 73:
Figure 74:
Figure 75:
Figure 76:
Figure 77:
Figure 78:
Figure 79:
Figure 80:
Annual Load Statistics by Zone ................................................................................. 56
Load Duration Curve – All hours ............................................................................... 57
Load Duration Curve – Top five percent of hours ..................................................... 58
Installed Capacity by Technology for each Zone ....................................................... 59
Installed Capacity by Type: 2007 to 2012.................................................................. 60
Annual Generation Mix .............................................................................................. 61
Marginal Unit Frequency by Fuel Type ..................................................................... 62
Average Wind Production .......................................................................................... 63
Summer Wind Production vs. Load ........................................................................... 64
Summer Renewable Production ................................................................................. 65
Wind Production and Curtailment .............................................................................. 66
Net Load Duration Curves ......................................................................................... 67
Top and Bottom Ten Percent of Net Load ................................................................. 68
Excess On-Line and Quick Start Capacity ................................................................. 69
Load Forecast Error .................................................................................................... 70
Frequency of Reliability Unit Commitments ............................................................. 71
Reliability Unit Commitment Capacity...................................................................... 72
Estimated Net Revenue by Zone and Unit Type ........................................................ 75
Gas Turbine Net Revenues ......................................................................................... 76
Combined Cycle Net Revenues .................................................................................. 77
Comparison of Net Revenue of Gas-Fired Generation between Markets .................. 79
Peaker Net Margin...................................................................................................... 81
Power Balance Penalty Curves................................................................................... 85
Projected Reserve Margins ......................................................................................... 90
Daily Average of Responsive Reserves provided by Load Resources ...................... 92
Pricing During Load Deployments............................................................................. 93
Residual Demand Index ............................................................................................. 96
Pivotal Supplier Frequency by Load Level ................................................................ 97
Ramp-Constrained Residual Demand Index .............................................................. 98
Frequency of Ramp Constrained Pivotal Supplier by Load Level ............................. 99
Surplus Capacity....................................................................................................... 101
Reductions in Installed Capability ........................................................................... 104
Short-Term Outages and Deratings .......................................................................... 105
Outages and Deratings by Load Level and Participant Size .................................... 106
Incremental Output Gap by Load Level and Participant Size – Step 1 .................... 108
Incremental Output Gap by Load Level and Participant Size – Step 2 .................... 109
LIST OF TABLES
Table 1:
Table 2:
Table 3:
Table 4:
15 Minute Price Changes as a Percent of Annual Average Price .................................. 14
Ancillary Service Deficiency......................................................................................... 34
Irresolvable Constraints ................................................................................................. 42
Power Balance Penalty Curve ....................................................................................... 86
ERCOT 2012 State of the Market Report
ERCOT 2012 State of the Market Report
Executive Summary
EXECUTIVE SUMMARY
A.
Introduction
This report reviews and evaluates the outcomes of the ERCOT wholesale electricity markets in
2012, and is submitted to the Public Utility Commission of Texas (“PUCT”) and the Electric
Reliability Council of Texas (“ERCOT”) pursuant to the requirement in Section 39.1515(h) of
the Public Utility Regulatory Act. It includes assessments of the incentives provided by the
current market rules and procedures, and analyses of the conduct of market participants. This
report also assesses the effectiveness of the Scarcity Pricing Mechanism (“SPM”) pursuant to the
provisions of PUCT Substantive Rule 25.505(g).
Key findings and statistics from 2012 include the following:
•
The ERCOT wholesale market performed competitively in 2012.
•
The ERCOT-wide load-weighted average real-time energy price was $28.33 per MWh in
2012, a 47 percent decrease from $53.23 per MWh in 2011. The decrease was primarily
driven by more moderate weather and much lower natural gas prices in 2012.
-
The average price for natural gas was 31 percent lower in 2012 than in 2011,
decreasing from $3.94 per MMBtu in 2011 to $2.71 per MMBtu in 2012.
-
After the extremes of 2011, loads in 2012 were more moderate with reduced
occurrences of shortage conditions. Total ERCOT load in 2012 was 2.7 percent
lower than 2011. Peak load decreased by 2.6 percent. Prices at the system-wide offer
cap were experienced in dispatch intervals which totaled 1.5 hours in 2012.
•
The total congestion revenue generated by the ERCOT real-time market in 2012 was
$480 million, a decrease of 9 percent from 2011. This decrease is not as large as might
be expected given the much larger decreases in average natural gas prices and real-time
energy prices.
-
The Odessa North transformer constraint was the most frequently occurring
transmission constraint in 2012. This, and other related localized constraints in west
Page i
Executive Summary
ERCOT 2012 State of the Market Report
Texas had significant financial impacts, causing the West zone average price to be
higher than the ERCOT average for the first time.
•
Even with the increased system-wide offer cap implemented in 2012, net revenues
provided by the market were at historic lows as energy prices fell substantially.
-
Net revenues were insufficient to support investment in new generation even though
planning reserve margins have fallen to levels that are close to the minimum planning
reserve targets.
-
These results underscore the importance of the resource adequacy issues currently
under consideration by the Commission, which we discuss in this report.
B.
Review of Real-Time Market Outcomes
As is typical in other wholesale electricity markets, only a small share of the power produced in
ERCOT is transacted in the spot market. However, prices in the real-time energy market are
very important because they set the expectations for prices in the forward markets where most
transactions take place. Unless there are barriers preventing arbitrage of the prices between the
spot and forward markets, the prices in the forward market should be directly related to the
prices in the spot market.
The average real-time energy prices by zone in 2009 through 2012 are shown below:
Average Real-Time Electricity Price
($ per MWh)
2009
2010
2011
2012
Page ii
ERCOT
Houston
North
South
West
$34.03
$34.76
$32.28
$37.13
$27.18
$39.40
$39.98
$40.72
$40.56
$33.76
$53.23
$52.40
$54.24
$54.32
$46.87
$28.33
$27.04
$27.57
$27.86
$38.24
Natural Gas
($/MMBtu)
$3.74
$4.34
$3.94
$2.71
ERCOT 2012 State of the Market Report
Executive Summary
The largest component of the all-in cost of wholesale electricity is the energy cost, which is
reflected by the locational marginal prices determined in the real-time energy market. ERCOT
average real-time market prices were 47 percent lower in 2012 than in 2011. The ERCOT-wide
load-weighted average price was $28.33 per MWh in 2012 compared to $53.23 per MWh in
2011.
Average All-in Price for Electricity in ERCOT
$160
Uplift
Ancillary Services
Energy
Natural Gas Price
$8
$7
$120
$6
$100
$5
$80
$4
$60
$3
$40
$2
$20
$1
$-
$-
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Electricity Price ($ per MWH)
$140
$9
Natural Gas Price ($ per MMBtu)
$180
2009
2010
2011
2012
The decrease in real-time energy prices was correlated with much lower fuel prices in 2012. The
steady decline in natural gas prices from June 2011 to April 2012 resulted in the 2012 average
natural gas price of $2.71 per MMBtu, a 31 percent decrease compared to $3.94 per MMBtu in
2011.
To depict how real-time energy prices vary by hour in each zone, the next figure shows the
hourly average price duration curve in 2012 for four ERCOT load zones. The Houston, North
and South load zones had similar prices over the majority of hours. The price duration curve for
the West zone is noticeably different than the other zones, with more hours with prices greater
Page iii
Executive Summary
ERCOT 2012 State of the Market Report
than $50 per MWh and more than 500 hours (6 percent of the time) when the average hourly
price was less than zero.
Zonal Price Duration Curves
$300
Houston
North
South
West
Electricity Price ($ per MWh)
$250
$200
< $0
3
6
3
571
$0-$50
8610
8573
8555
6741
Frequency of Prices
$50-$100
$100-$200
91
49
125
49
128
65
1103
191
> $200
31
31
33
178
$150
Houston
North
South
West
$100
$50
$0
($50)
0
1,000
2,000
3,000
4,000
5,000
Number of Hours
6,000
7,000
8,000
As observed over the past few years, West zone prices are lower than the rest of ERCOT when
high wind output in the west results in congested transmission interfaces from the West zone to
the other zones in ERCOT. Recently, prices higher than the rest of ERCOT have occurred in the
West zone due to local transmission constraints that typically occur under low wind and high
load, or outage conditions. Specifics about these transmission constraints are provided in
Section III, Transmission and Congestion.
After the extremes of 2011, weather conditions in Texas returned to closer to normal in 2012.
As more fully discussed in Section IV, Load and Generation, overall demand for electricity was
lower in 2012 than in 2011, resulting in much fewer occasions when the available supply
generation capacity was unable to meet customer demands. This resulted in a decreased
likelihood that the available generation capacity was not sufficient to meet customer demands for
electricity and maintain the required reliability reserves.
Page iv
ERCOT 2012 State of the Market Report
Executive Summary
As more fully described later in Section V, Resource Adequacy, the nodal market causes energy
prices to rise toward the system-wide offer cap as available operating reserves approach
minimum required levels to reflect the degradation in system reliability.
Duration of Prices at the System-Wide Offer Cap
20
17.4
15
Hours
Prices were at the System Wide Offer Cap for
28.4 hours in 2011 and
1.5 hours in 2012
10
6.0
5
2011
Dec
Nov
Oct
Sep
Aug
0.1
Jul
Jun
May
0.4
Apr
Mar
Feb
Jan
Dec
Oct
Sep
1.0
0.7
0.2
Nov
0.2
Aug
1.1
Jul
0.6
Jun
Mar
Feb
Jan
0
0.1
May
0.6
Apr
1.5
2012
Presented in the figure above is the aggregated amount of time represented by all five-minute
dispatch intervals where the real-time energy price was at the system-wide offer cap, displayed
by month. Prices during 2012 were at the system-wide offer cap for only 1.5 hours, a significant
reduction from the 28.4 hours experienced in 2011. Approved during 2012, PUCT SUBST.
R.25.508 increased the system-wide offer cap to $4,500 per MWh effective August 1, 2012. As
shown in the figure above, there was only a brief period when energy prices rose to the cap after
this change was implemented.
Finally, after the multiple protocol revisions implemented in 2012, the non-spinning reserve
deployment process remains sub-optimal from a reliability and efficiency perspective. As more
fully described in Section I.H, Recommendations, we continue to recommend that ERCOT
Page v
Executive Summary
ERCOT 2012 State of the Market Report
develop a mechanism that will rationally commit generation and load resources that can start or
curtail within 30 minutes. This deficiency in ERCOT’s nodal market design should be addressed
by implementing a “look ahead” dispatch functionality for the real-time market to produce an
energy and ancillary services commitment and dispatch results that are co-optimized and
recognize anticipated changes in system demands. This additional functionality represents a
major change to ERCOT systems; one we recommend together with improved pricing provisions
that will allow offers from load resources to set prices if they are required to meet system
demand.
C.
Review of Day-Ahead Market Outcomes
ERCOT’s centralized day-ahead market allows participants to make financially binding forward
purchases and sales of power for delivery in real-time. Although all bids and offers are
evaluated in the context of their ability to reliably flow on the transmission network, there are no
operational obligations resulting from the day-ahead market clearing. These transactions are
made for a variety of reasons, including satisfying the participant’s own supply, managing risk
by hedging the participant’s exposure to the real-time market, or arbitraging with the real-time
markets. For example, load serving entities can insure against volatility in the real-time market
by purchasing in the day-ahead market. Finally, the day-ahead market plays a critical role in
coordinating generator commitments. For all of these reasons, the performance of the day-ahead
market is essential.
Day-ahead market performance is primarily evaluated by the degree to which its outcomes
converge with those of the real-time market because the real-time market reflects actual physical
supply and demand for electricity. In a well-functioning market, participants should eliminate
sustained price differences on a risk-adjusted basis by making day-ahead purchases or sales to
arbitrage them over the long-term.
The figure below shows the price convergence between the day-ahead and real-time market,
summarized by month. The simple average of day-ahead prices in 2012 was $29 per MWh,
compared to the simple average of $27 per MWh for real-time prices. The average absolute
difference between day-ahead and real-time prices was $9.96 per MWh in 2012; much lower
Page vi
ERCOT 2012 State of the Market Report
Executive Summary
than in 2011 when average of the absolute difference was $24.50 per MWh. This reduction was
due to fewer occurrences of shortage intervals and associated high prices in 2012.
Convergence between Forward and Real-Time Energy Prices
60
50
$ per MWh
40
2009
2010
2011
2012
Average
Average
Day-Ahead Real Time
Price
Price
$38
$35
$42
$40
$46
$43
$29
$27
Day Ahead
Real Time
Absolute Difference
30
20
10
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
This day-ahead premium is consistent with expectations due to the much higher volatility of realtime prices. Risk is lower for loads purchasing in the day-ahead and higher for generators selling
day-ahead. The higher risk for generators is associated with the potential of incurring a forced
outage and as a result, having to buy back energy at real-time prices. This explains why the
highest premiums tend to occur during the months with the highest demand and highest prices.
Overall, the day-ahead premiums were very similar to the differences observed in 2010 and
2011, but remain higher than observed in other organized electricity markets. Real-time energy
prices in ERCOT are allowed to rise to levels that are much higher than what is allowed in other
organized electricity markets, which increases risk and helps to explain the higher day-ahead
premiums regularly observed in ERCOT. Although most months experienced a day-ahead
premium (e.g., $10 per MWh in August), it should not be expected over time that every month
Page vii
Executive Summary
ERCOT 2012 State of the Market Report
will always produce a day-ahead premium as the real-time risks that lead to the premiums will
materialize unexpectedly on occasion, resulting in real-time prices that exceed day-ahead prices
(e.g., in March).
Summarized in the figure below is the volume of day-ahead market activity by month. It shows
that day-ahead purchases are approximately 45 percent of real-time load.
Volume of Day-Ahead Market Activity by Month
70
Day Ahead Market Volume (GW)
60
Three Part Awards
Energy Only Awards
Load
PTP Obligations
Day-Ahead Purchase
50
40
30
20
10
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Point to Point Obligations are financial instruments purchased in the day-ahead market.
Although they do not provide any energy supply themselves, they do provide the ability to avoid
the congestion costs associated with transferring the delivery of energy from one location to
another. To provide a volume comparison we aggregate all of these “transfers”, netting location
specific injections against withdrawals. By adding the aggregated transfer capacity associated
with purchases of PTP Obligations, we find that on average total volumes transacted in the dayahead market are greater than real-time load.
Page viii
ERCOT 2012 State of the Market Report
Executive Summary
Ancillary Service capacity is procured as part of the day-ahead market clearing. The figure
below shows the monthly total ancillary service costs per MWh of ERCOT load and the average
real-time energy price for 2009 through 2012. Total ancillary service costs are generally
correlated with real-time energy price movements, which in turn are highly correlated with
natural gas price movements. The average ancillary service cost per MWh of load decreased to
$1.06 per MWh in 2012 compared to $2.41 per MWh in 2011, a decrease of 56 percent. Total
ancillary service costs decreased from 4.5 percent of the load-weighted average energy price in
2011 to 3.7 percent in 2012.
Ancillary Service Costs per MWh of Load
$10
Average Price of 2009 2010 2011 2012
Ancillary Services $1.15 $1.26 $2.41 $1.06
$144
$7
$6
$128
Real Time Electricity Price ($ per MWh)
$8
$112
Real Time Electricity Price
Ancillary Service Prices
$96
$5
$80
$4
$64
$3
$48
$2
$32
$1
$16
$0
$0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Ancillary Service Price ($ per MWh of Load)
$9
$160
2009
D.
2010
2011
2012
Transmission and Congestion
There was a marked change in the nature of real-time transmission congestion during 2012 when
compared to previous years. For the past several years a significant portion of real-time
transmission congestion could be described as limiting the export of wind generation from the
West zone to the load centers across the rest of ERCOT. Transmission congestion in 2012 was
Page ix
Executive Summary
ERCOT 2012 State of the Market Report
more significantly the result of limitations on the ability to get generation to loads in the West
zone. Some portion of the limitation can be attributed to transmission outages taken to enable
the construction of new CREZ transmission lines. Another factor is that loads in far west Texas
have increased more than the system-wide average.
The total congestion revenue generated by the ERCOT real-time market in 2012 was
$480 million, a decrease of 9 percent from 2011. This decrease is not as large as might be
expected given the much larger decreases in average natural gas prices real-time energy prices.
Two factors influencing the overall costs of congestion in 2012 were the significant financial
impact of several localized transmission constraints in far west Texas and the higher frequency
of active transmission constraints.
Shown below is a comparison of the amount of time transmission constraints were active at
various load levels in 2012 and 2011.
Frequency of Active Constraints
100%
90%
2011
2012
80%
Percent of Time
70%
60%
50%
40%
30%
20%
10%
0%
<25
Page x
25-30 30-35 35-40 40-45 45-50 50-55 55-60 60-65
Load (GW)
>65
Annual
ERCOT 2012 State of the Market Report
Executive Summary
We observe that in 2012 the likelihood of having an active transmission constraint was higher
than it was in 2011 and that for loads above 45 GW the frequency was much higher. During
2011, we observed that at higher system loads ERCOT operators did not always activate (or
sometimes de-activated) transmission constraints. This was due to a concern that by having a
constraint active during periods of high demand the total capacity available to serve load may be
limited. However, ERCOT’s dispatch software contains parameters that allow it to automatically
make the correct decision about when to violate transmission constraints when necessary to serve
total system load. Therefore, ERCOT modified their practice in 2012 to retain active
transmission constraints even during periods of high demand.
The figure below displays the ten most costly real-time constraints and indicates that the Odessa
North 138/69 kV transformer constraint was by far the most highly valued during 2012. This
constraint became more pronounced from 2011 to 2012 and is mainly attributed to load growth
in far west Texas driven by increased oil and natural gas activity.
Top Ten Real-Time Constraints
The Odessa North 138/69 kV transformer typically overloads during low wind conditions. The
characteristics that load the Odessa North 138/69 kV transformer are the same conditions that
Page xi
Executive Summary
ERCOT 2012 State of the Market Report
also affect Odessa to Odessa North 138 kV line, which is shown as the seventh constraint on the
list. Not only did this constraint have nearly twice the financial impact of the second constraint
on the list, its impact was more than 40 percent greater than the top constraint from 2011. Its
magnitude is even more significant given the overall lower costs of energy in 2012.
Not surprisingly, much public attention was focused on this constraint; much of it questioning its
causes and the potential for short-term remedies. ERCOT and the local transmission provider
were able to identify two transmission lines, which when opened greatly reduced congestion
around the Odessa North station without causing other reliability concerns. After the lines were
opened in mid-September, congestion around Odessa North was almost eliminated for the rest of
the year.
The figure below presents a summary of the utilization of the West to North interface. The West
to North constraint continued to be active at some point during every month of 2012 but with far
less impact than in 2011.
Utilization of the West to North Interface Constraint
3500
Percent Utilization
100%
90%
3000
80%
2500
70%
60%
MW
2000
Physical Limit
1500
50%
40%
Actual Flow
1000
30%
20%
500
0
10%
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
2011
Page xii
2012
0%
ERCOT 2012 State of the Market Report
Executive Summary
Through the years this constraint has been a major impediment to delivering all the wind
generation located and produced in the western reaches of ERCOT to the load centers. This
constraint had the highest financial impact of all real-time transmission constraints during 2011,
but in 2012 its impact dropped to one third that level. The reduction was a result of the
combined impact of higher loads in the west and increased transfer capability due to the first
CREZ transmission lines being placed in service.
Through July 2012, the average physical limit was approximately 2,200 MW and the average
actual flow during constrained intervals was approximately 1,900 MW. After July 2012, the
physical limit increased to an average of 2,900 MW and the actual flow increased to
approximately 2,500 MW. In March 2012, a new real-time analysis tool was implemented to
better track the dynamic nature of the transient stability limit of the West to North interface.
However, there was not a noticeable increase to the transfer limit corresponding to its
implementation due to the effects of transmission outages occurring to accommodate
maintenance activities and the installation of CREZ lines. Many of these outages were complete
by July 2012 which accounts for the increase in the West to North limit.
The average annual utilization of the West to North constraint was 87 percent in 2012, which
compares favorably to 78 percent utilization experienced in 2011. Over the long term, the
physical limit will continue to increase as CREZ transmission projects are completed.
E.
Load and Generation
The figure below shows peak load and average load in each of the ERCOT zones from 2009 to
2012. In each zone, as in most electrical systems, peak demand significantly exceeds average
demand. The North zone is the largest zone (with about 38 percent of the total ERCOT load);
the South and Houston zones are comparable (27 percent); while the West zone is the smallest
(8 percent of the total ERCOT load). The figure also shows the annual non-coincident peak load
for each zone. This is the highest load that occurred in a particular zone for one hour during the
year; however, the peak can occur in different hours for different zones. As a result, the sum of
the non-coincident peaks for the zones was greater than the annual ERCOT peak load.
Page xiii
Executive Summary
ERCOT 2012 State of the Market Report
Annual Load Statistics by Zone
35
Real-Time Peak Load
Change in Real-Time Load
(2011 to 2012)
Peak
Average
ERCOT
-2.6%
-3.0%
Average Real-Time Load
30
Houston
North
South
West
Load (GW)
25
20
-4.7%
-3.8%
0.3%
2.7%
-2.2%
-4.3%
-3.1%
1.8%
15
10
5
Houston
North
South
2012
2011
2010
2009
2012
2011
2010
2009
2012
2011
2010
2009
2012
2011
2010
2009
0
West
Total ERCOT load decreased from 334 TWh in 2011 to 325 TWh in 2012, a decrease of
2.7 percent or an average of 1,130 MW every hour. Similarly, the ERCOT coincident peak
hourly demand decreased from 68,311 MW in 2011 to 66,559 MW, a decrease of 1,752 MW, or
2.6 percent. The results at the zonal level are not consistent. Average load decreased in three of
the four zones, but grew by 1.8 percent in the West zone.
New generation resources in 2012 totaled approximately 1 GW; most of which were wind units
and the remainder were solar and biomass. Comparing the current mix of installed generation
capacity to that in 2007, we find that over these six years wind and coal generation are the two
categories with the most increased capacity. The sizable additions in these two categories have
been more than offset by retirements of natural gas-fired steam units, resulting in less installed
capacity in 2012 than there was in 2007.
Shown below is ERCOT’s most current projection of reserve margins. It indicates that the
region will have a 13.2 percent reserve margin heading into the summer of 2013. With the
Page xiv
ERCOT 2012 State of the Market Report
Executive Summary
addition of recently announced generation additions, in 2014 the reserve margin is expected to
reach 13.8 percent -- just barely above the current target. The bulk of the new capacity being
added is natural gas-fired generation, approximately a quarter of which are expansions at existing
facilities.
Projected Reserve Margins
16%
Existing Capacity
New Coal
New Gas
New Other
New Wind (8.7% of installed capacity)
Target Reserve Margin: 13.75%
14%
Projected Reserve Margin
12%
10%
8%
6%
4%
2%
0%
2013
2014
2015
2016
2017
2018
Source: ERCOT Capacity Demand Reserve Reports / 2013 data from Winter 2012, 2014 - 2018 from May 2013
The figure below shows the percent of annual generation from each fuel type for the years 2008
through 2012. The generation share from wind has increased every year, reaching 9 percent of
the annual generation requirement in 2012, up from 5 percent in 2008. During the same period,
the percentage of generation provided by natural gas decreased from 43 percent in 2008 to
38 percent in 2010, before increasing to 45 percent in 2012. Correspondingly, the percentage of
generation produced by coal units increased from 37 percent to 40 percent in 2010 before
decreasing to 34 percent in 2012. The increase in the share of generation produced by natural
gas, and corresponding reduction in coal generation is due to historically low price of natural gas
in 2012.
Page xv
Executive Summary
ERCOT 2012 State of the Market Report
Annual Generation Mix
100%
90%
80%
Annual Generation Mix
70%
Other
60%
Hydro
Natural Gas
50%
Wind
40%
Coal
Nuclear
30%
20%
10%
0%
2008
2009
2010
2011
2012
While coal/lignite and nuclear plants operate primarily as base load units in ERCOT, it is the
reliance on natural gas resources that drives the high correlation between real-time energy prices
and the price of natural gas fuel. There is approximately 23 GW of coal and nuclear generation
in ERCOT. Generally, when ERCOT load is above this level, natural gas resources will be on
the margin and set the real-time energy spot price. However, due to the low price of natural gas
in 2012, we observe that the share of generation produced from coal-fired and nuclear units
decreased to less than half of the energy in ERCOT, with the reduction coming from decreased
coal generation.
Increasing levels of wind resource in ERCOT also has important implications for the net load
duration curve faced by the non-wind fleet of resources. Net load is defined as the system load
minus wind production. The figure below shows the net load duration curves for selected years
since 2007, normalized as a percent of peak load. This figure shows the continued erosion of
remaining energy available for non-wind units to serve during most hours of the year, with much
less impact during the highest loads.
Page xvi
ERCOT 2012 State of the Market Report
Executive Summary
Net Load Duration Curve
100%
90%
2012
Normalized Net Load
80%
2009
70%
2007
60%
50%
40%
30%
20%
0
1,000
2,000
3,000
5,000
4,000
Hours
6,000
7,000
8,000
Thus, although the peak net load and reserve margin requirements are projected to continue to
increase and create an increasing need for non-wind capacity to meet net load and reliability
requirements, the non-wind fleet is expected to operate for fewer hours as wind penetration
continues to increase. This outlook further reinforces the importance of efficient energy pricing
during peak demand conditions and other times of system stress, particularly within the context
of the ERCOT energy-only market design.
Even with the increased development activity in the coastal area of the South zone, more than
80 percent of the wind resources in the ERCOT region are located in west Texas. The wind
profiles in this area are such that most of the wind production occurs during off-peak hours or
other times of relatively low system demand. This profile results in only modest reductions of
the net load relative to the actual load during the hours of highest demand, but much more
significant reductions in the net load relative to the actual load in the other hours of the year.
The trend shown from 2007 in the figure above may continue with the addition of new wind
Page xvii
Executive Summary
ERCOT 2012 State of the Market Report
resources and the reduction in the curtailment of existing wind resources after completion of new
transmission facilities.
The next figure compares the output during the summer months of June through August from
wind units located in the coastal area of the South zone with those located elsewhere in ERCOT.
Summer Wind Production vs. Load
3,500
Coastal
Non-Coastal
Average Wind Production (MW)
3,000
2,500
2,000
1,500
1,000
500
0
<35
35-40
40-45
45-50
50-55
ERCOT Load (GW)
55-60
60-65
>65
It shows a strong negative relationship between non-coastal wind output and increasing load
levels. It further shows that the output from wind generators located in the coastal area of the
South zone is much more highly correlated with peak electricity demand.
F.
Resource Adequacy
Long-Term Economic Signals: Net Revenue
One of the primary functions of the wholesale electricity market is to provide economic signals
that will encourage the investment needed to maintain a set of resources that are adequate to
satisfy the system’s demands and reliability needs. These economic signals are evaluated by
Page xviii
ERCOT 2012 State of the Market Report
Executive Summary
estimating the “net revenue” new resources would receive from the markets. Net revenue is the
total revenue that can be earned by a new generating unit less its variable production costs. Put
another way, it is the revenue in excess of short-run operating costs that is available to recover a
unit’s fixed and capital costs, including a return on the investment. Net revenues from the
energy and ancillary services markets together provide the economic signals that inform
suppliers’ decisions to invest in new generation or retire existing generation. In long-run
equilibrium, markets should provide sufficient net revenue to allow an investor to receive a
return of, and on an investment in a new generating unit.
Estimated Net Revenue
$300
Regulation
Reserves
Energy Sales
Net Revenue ($ per kW-yr)
$250
$200
$150
$100
$50
CC
CT
Coal
Nuke
2011
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
$-
CC
CT
Coal
Nuke
2012
The figure above shows the results of the net revenue analysis for four types of hypothetical new
units in 2011 and 2012. These are: (a) natural gas-fired combined-cycle, (b) natural gas-fired
combustion turbine, (c) coal-fired generator, and (d) a nuclear unit. For the natural gas-fired
technologies, net revenue is calculated by assuming the unit will produce energy in any hour for
which it is profitable and by assuming it will be available to sell reserves and regulation in other
hours that it is available. For coal and nuclear technologies, net revenue is calculated by
assuming that the unit will produce at full output.
Page xix
Executive Summary
ERCOT 2012 State of the Market Report
The energy net revenues are computed based on the generation-weighted settlement point prices
from the real-time energy market. Weighting the energy values in this way masks what may be
very high locational values for a specific generator location. Some generators may also receive
uplift payments because of their specific reliability contributions, either as a reliability must run,
or through the reliability unit commitment. This source of revenue is not considered in this
analysis. The analysis also includes simplifying assumptions that can lead to over-estimates of
the profitability of operating in the wholesale market. Start-up costs and minimum run times and
ramp restriction, which can prevent the natural gas generators from profiting during brief price
spikes, are not explicitly accounted for in the net revenue analysis. Despite these limitations, the
net revenue analysis provides a useful summary of signals for investment in the wholesale
market.
The figure above shows that the net revenue for every generation technology type decreased
substantially in 2012 compared to each zone in 2011.
•
For a new coal-fired unit, the estimated net revenue requirement is approximately $210 to
$270 per kW-year. The estimated net revenue in 2012 for a new coal unit was
approximately $35 per kW-year.
•
For a new nuclear unit, the estimated net revenue requirement is approximately $280 to
$390 per kW-year. The estimated net revenue in 2012 for a new nuclear unit was
approximately $134 per kW-year.
•
For a new natural gas-fired combustion turbine, the estimated net revenue requirement is
approximately $80 to $105 per kW-year. The estimated net revenue in 2012 for a new
gas turbine was approximately $25 per kW-year.
•
For a new combined cycle unit, the estimated net revenue requirement is approximately
$105 to $135 per kW-year. The estimated net revenue in 2012 for a new combined cycle
unit was approximately $42 per kW-year.
These results indicate that the ERCOT markets would not have provided sufficient revenues to
support profitable investment in any of the types of generation technology evaluated. Higher
energy prices in the West zone during 2012 resulted in higher net revenues in that zone, but they
were still not high enough to support new entry there. The net revenues in 2012 were much
lower than in 2011. However, it is important to recognize that 2011 was highly anomalous, with
Page xx
ERCOT 2012 State of the Market Report
Executive Summary
some of the hottest summer temperatures on record. Net revenues may have been sufficient to
cover the costs of a new combined cycle or new combustion turbine in 2011, however, we would
not expect this to be consistently true in years with comparable reserve margins absent the
extreme weather conditions, as evidenced by the 2012 net revenue results.
Shortage Pricing, Capacity Markets, and Resource Adequacy
Efficient electricity markets allow energy prices to rise substantially at times when the available
supply is insufficient to simultaneously meet both energy and minimum operating reserve
requirements. Ideally, energy and reserve prices during shortages should reflect the diminished
system reliability under these conditions, which is equal to the increased probability of “losing”
load times the expected value of the lost load. Allowing energy prices to rise during shortages
mirrors the outcome expected if loads were able to actively specify the quantity of electricity
they wanted and the price they would be willing to pay. The energy-only market design relies
exclusively on these relatively infrequent occurrences of high prices to provide the appropriate
price signal for demand response and new investment when required. In this way, energy-only
markets can provide price signals that will sustain a portfolio of resources to be used in real time
to satisfy the needs of the system. However, this portfolio may produce a planning reserve
margin that is less than the planning reserve target.
The nodal market implementation brought about more reliable and efficient shortage pricing.
Modifications implemented during 2012, which introduced offer floors associated with the
deployment of generator-provided responsive reserves and non-spinning reserves, further
improved pricing outcomes. However, ERCOT’s real-time energy pricing can be improved by
ensuring the value of curtailed load is fully reflected in prices when load resources are deployed
and further improving its shortage pricing as recommended in this report.
The PUCT has devoted considerable effort over the past two years deliberating issues related to
resource adequacy. These deliberations have resulted in changes to the rules governing the
system-wide offer cap and the peaker net margin (“PNM”) mechanism. The system-wide offer
cap was increased to $4,500 per MWh effective August 1, 2012 and is scheduled to increase
every year up to $9,000 per MWh on June 1, 2015. This is intended to raise market revenues to
help address resource adequacy concerns. However, inflating the system-wide offer cap may
Page xxi
Executive Summary
ERCOT 2012 State of the Market Report
raise efficiency concerns if the resulting energy prices are set at or above levels consistent with
the value of lost load during periods when the system is only slightly short of operating reserves
and involuntary load curtailment is not imminent. This is a concern because setting prices
substantially higher than the expected value of lost load can cause market participants to take
inefficient actions, resulting in higher overall market costs. To address this concern, we
recommend that ERCOT:
•
Modify the slope of the existing power balance penalty curve and the offer floor for
responsive reserve service to provide a more gradual slope up to the system-wide offer
cap; and
•
Modify the automatic pricing of unoffered capacity such that it is not all priced at the
system-wide offer cap to avoid the inefficiencies associated with the automated economic
withholding of such capacity.
Regardless of the means by which revenues are produced in a wholesale electricity market, it is
fundamental that investment will only occur when the total net revenues expected by the investor
are greater than its entry costs (including profit on its investment). Additionally, these sources of
revenue must be available to all resources, both new and existing, in order to facilitate efficient
investment, maintenance, and retirement decisions by all suppliers.
In an energy only market, the primary source of such revenue is the net revenues received during
periods of shortage. Expectations about both the magnitude of the energy price during shortage
conditions and the frequency of shortage conditions are the primary means to attract new
investment in an energy-only market. If the expected revenues are not high enough to facilitate
enough investment to satisfy the planning reserve target, one option is to increase the shortage
pricing levels to levels that substantially exceed the expected value of lost load. As the planning
reserve levels grow, however, the frequency of shortages will tend to drop sharply, which can
make it difficult to use this means to meet planning reserve targets. 1 Additionally, as discussed
below, such approaches introduce costly operational inefficiencies into the ERCOT energy
markets.
1
The difficulty of relying primarily on shortage pricing will depend on how high the planning reserve target
is relative to the planning reserve levels any energy-only market priced at the expected value of lost load
would provide. See the discussion of the Brattle Report below.
Page xxii
ERCOT 2012 State of the Market Report
Executive Summary
Most other competitive electricity markets do not rely solely on shortage pricing to generate
sufficient revenue to support the capacity additions necessary to satisfy their planning reserve
requirements. They employ capacity markets to competitively generate capacity payments over
the year that are made to suppliers in return for meeting defined capacity obligations. Capacity
prices and associated payments vary monthly or annually based on long-term planning reserve
levels, independent of the real-time supply and demand conditions. These capacity markets are
designed to ensure that a specified planning reserve margin is achieved.
In 2012, ERCOT engaged The Brattle Group to assess its resource adequacy outlook by
evaluating a number of market design scenarios. 2 Brattle also supplemented its report with a
comparison of costs and reliability for the energy-only market and two capacity market scenarios
with 10% and 14% reserve margin requirements. 3 The results of this analysis are summarized in
the table below.
2
ERCOT Investment Incentives and Resource Adequacy, The Brattle Group, PUCT Docket No. 37987 (June 1,
2012).
3
Customer Cost Comparison, The Brattle Group, PUCT Docket No. 40000 (Sept. 4, 2012).
Page xxiii
Executive Summary
ERCOT 2012 State of the Market Report
As indicated in the table above, Brattle estimates that even with $9,000 per MWh system-wide
offer caps, economic equilibrium for the ERCOT energy-only market is achieved at an 8 percent
planning reserve margin, although the actual reserve margin outcomes will be uncertain. Brattle
further estimates that in the energy-only market at annual equilibrium, wholesale generation
costs will be $18.3 billion. In contrast, Brattle’s assessment of a capacity market with a more
certain 14 percent reserve margin expectation, estimates generation costs at annual equilibrium
to be $18.7 billion.
It is important to recognize that this increase in cost is not due to the introduction of the capacity
market, it is due to the requirement to sustain a planning reserve margin greater than 8 percent.
In fact, the Brattle analysis indicates that a capacity market would deliver the higher planning
reserve margin at a relatively low incremental cost with much more certainty. Further, recent
studies have indicated that to maintain the same small level of risk of having an involuntary
curtailment of firm load, the planning reserve target should be increased from 13.75 percent to
approximately 16 percent. Hence, the difficulty of satisfying ERCOT’s planning needs with
shortage pricing alone will grow if this recommendation is adopted.
In response to these observations, proposals have been put forth that would introduce significant
operational inefficiencies into the ERCOT energy markets, such as a requirement to substantially
increase the quantity of operating reserves ERCOT procures and to, by rule, economically
withhold these surplus reserves from the market. Such approaches would introduce significant
inefficiencies into ERCOT day ahead and real time operations in an effort to manufacture more
frequent shortage pricing and a higher planning reserve margin than would be achieved in a pure
energy-only market framework. However, such approaches will not guarantee that the planning
reserve targets will be satisfied and, because of the resulting operational inefficiencies, will be
more costly for ERCOT’s consumers. Hence, consistent with Brattle’s findings, it is our view
that if the planning reserve margin is viewed as a minimum requirement, implementation of a
capacity market is the most efficient mechanism to achieve this objective. As observed by
Brattle, a well-designed capacity market can efficiently meet a planning reserve requirement
without impairing the efficiency of energy market operations. However, there are many
Page xxiv
ERCOT 2012 State of the Market Report
Executive Summary
determinations required in the design, implementation and maintenance of a capacity market
construct. 4
G.
Analysis of Competitive Performance
The report evaluates market power from two perspectives, structural (does market power exist)
and behavioral (have attempts been made to exercise it). The Residual Demand Index (“RDI”) is
used to as the primary indicator of potential structural market power. The RDI measures the
percentage of load that cannot be served without the resources of the largest supplier, assuming
that the market could call upon all committed and quick-start capacity owned by other suppliers.
When the RDI is greater than zero the largest supplier is pivotal; that is, its resources are needed
to satisfy the market demand. When the RDI is less than zero, no single supplier’s resources are
required to serve the load as long as the resources of its competitors are available.
The RDI is a useful structural indicator of potential market power, although it is important to
recognize its limitations. As a structural indicator, it does not illuminate actual supplier behavior
to indicate whether a supplier may have exercised market power. The RDI also does not indicate
whether it would have been profitable for a pivotal supplier to exercise market power. However,
it does identify conditions under which a supplier would have the ability to raise prices
significantly by withholding resources.
4
ERCOT Investment Incentives and Resource Adequacy, The Brattle Group, at 115-119, PUCT Docket No. 37987
(June 1, 2012).
Page xxv
Executive Summary
ERCOT 2012 State of the Market Report
Pivotal Supplier Frequency by Load Level
50%
100%
90%
Percent of Hours with Pivotal Supplier
Percent of Hours at Load Levels
40%
80%
Percent of Time at Load Level
35%
70%
30%
60%
25%
50%
20%
40%
15%
30%
10%
20%
5%
10%
0%
0%
<25
25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 60 - 65
Load (GW)
Percent of Hours with Pivotal Supplier
45%
>65
The figure above summarizes the results of the RDI analysis by displaying the percent of time at
each load level there was a pivotal supplier. At loads greater than 65 GW there was a pivotal
supplier 100 percent of the time. The figure also displays the percent of time each load level
occurs. Combining these values we find that there was a pivotal supplier in approximately
12 percent of all hours of 2012. As a comparison, the same system-wide measure for the
Midwest ISO showed less than 1 percent of all hours with a pivotal supplier.
The behavioral aspects of market power abuse are evaluated by calculating an “output gap.” The
output gap is defined as the quantity of energy that is not being produced by in-service capacity
even though the in-service capacity is economic by a substantial margin given the real-time
energy price. A participant can economically withhold resources, as measured by the output gap,
by raising its energy offers so as not to be dispatched.
Resources are considered for inclusion in the output gap when they are committed and producing
at less than full output. Energy not produced from committed resources is included in the output
Page xxvi
ERCOT 2012 State of the Market Report
Executive Summary
gap if the real-time energy price exceeds by at least $50 per MWh that unit’s mitigated offer cap,
which serves as an estimate of the marginal production cost of energy from that resource.
The output gap is measured at both steps in ERCOT’s two-step dispatch because if a market
participant has sufficient market power, it might raise its offer in such a way to increase the
reference price in the first step of ERCOT’s dispatch process. Although in the second step, the
offer appears to be mitigated, the market participant has still influenced the market price. This
output gap is measured by the difference between the capacity level on a generator’s original
offer curve at the first step reference price and the capacity level on the generator’s cost curve at
the first step reference price. However, this output gap is only indicative because no output
instructions are sent based on the first step. It is only used to screen out whether a market
participant is withholding in a manner that may influence the reference price.
The ultimate output gap is measured by the difference between a unit’s operating level and the
output level had the unit been competitively offered to the market. In the second step of the
dispatch, the after-mitigation offer curve is used to determine dispatch instructions and locational
prices. Even though the offer curve is mitigated there is still the potential for the mitigated offer
curve to be increased as a result of a high first step reference price due to a market participant
exerting market power.
Page xxvii
Executive Summary
ERCOT 2012 State of the Market Report
Incremental Output Gap by Load Level and Participant Size
Average Percent of Capacity
1.0%
0.8%
0.6%
0.4%
Small
Large
0.2%
0.0%
1.0%
Average Percent of Capacity
Step 1 Dispatch
Step 2 Dispatch
0.8%
0.6%
0.4%
0.2%
0.0%
<25 25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 60 - 65 >65
Load (GW)
The figure above shows the magnitude of the output gap to be very small, even at the highest
load levels, for both steps in the dispatch process, and for both small and large generators. These
small quantities raise no competitive concerns. In addition to this metric, we also evaluate
outages, deratings, and economic units that were not committed to identify other means suppliers
may have used to withhold resources. Based on the analysis above and our other monitoring
screens, we find that the ERCOT nodal wholesale market performed competitively in 2012.
Page xxviii
ERCOT 2012 State of the Market Report
H.
Executive Summary
Recommendations
Last year we recommended changes to the automated mitigation procedures that are part of the
real-time dispatch to eliminate the occurrences of over-mitigation we have observed. As more
fully described in Section I.E, Mitigation at page 20, we support the introduction of a test to
determine whether a unit is either contributing to, or helping to resolve a transmission constraint
and only subject the relieving units to mitigation. These changes were included in NPRR520 and
should substantially reduce the occurrence of mitigating resources that are not in a position to
exert market power related to the relief of transmission constraints. Various parameters will be
approved related to the implementation of these changes, currently scheduled for summer 2013,
and performance should be closely monitored to determine if any adjustments are required.
Last year we also recommended a change to the real-time market software to allow it to “look
ahead” a sufficient amount of time to better commit load and generation resources that can be
online within 30 minutes. More discussion of this topic can be found starting on page 84 in
Section V.B, Effectiveness of the Scarcity Pricing Mechanism. We still believe this functionality
would enhance the performance of the ERCOT market. However, this will need to be
coordinated with the other fundamental market design changes still currently under
consideration.
Whatever these future market design changes may entail, we recommend stakeholder
consideration of the following three modifications, particularly as the system-wide offer cap rises
above $5,000 per MWh.
1. Modify the slope of the existing power balance penalty curve and the offer floors for
responsive reserve service to provide a more gradual slope up to the system-wide offer
cap such that the cap is reached when operating reserves are down to the level that
ERCOT would initiate involuntary curtailment of firm load.
o If system-wide offer caps of $5,000, $7,000 or $9,000 are intended to reflect the
value of lost load, then real-time energy prices should only get to those levels
once firm load is being involuntarily curtailed.
Page xxix
Executive Summary
ERCOT 2012 State of the Market Report
o A more “well-behaved” reserve shortage pricing function could be achieved in the
following ways:

Implement real-time co-optimization of energy and reserves with an
operating reserve demand curve;

Introduce an operating reserve demand curve but not include real-time cooptimization.

Adjust the current operating reserve offer floors to better reflect the loss of
load probability and the value of lost load at various levels of operating
reserves.
2. Modify the Protocols related to proxy offer curve provisions such that all unoffered
capacity is not automatically priced at the system-wide offer cap. Currently, if available
capacity does not have an associated energy offer, ERCOT’s dispatch software “fills in”
with an offer priced at the system-wide offer cap. During 2012, the average amount of
capacity priced in this manner exceeded 100 MW.
3. Implement changes that ensure ERCOT deployments of load resources, Emergency
Response Service (ERS), or the involuntary curtailment of firm load are reflected in the
real-time dispatch energy and reserve prices. This may be achieved through various
means, either as a component of integrating load bids in the real-time dispatch software,
or through simple administrative shortage pricing rules.
Page xxx
ERCOT 2012 State of the Market Report
I.
A.
Real-Time Market
REVIEW OF REAL-TIME MARKET OUTCOMES
Real-Time Market Prices
Our first analysis evaluates the total cost of supplying energy to serve load in the ERCOT
wholesale market. In addition to the costs of energy, loads incur costs associated with ancillary
services and a variety of non-market based expenses referred to as “uplift”. We have calculated
an average “all-in” price of electricity for ERCOT that is intended to reflect wholesale energy
costs as well as these additional costs.
Figure 1: Average All-in Price for Electricity in ERCOT
$160
Uplift
Ancillary Services
Energy
Natural Gas Price
$8
$7
$120
$6
$100
$5
$80
$4
$60
$3
$40
$2
$20
$1
$-
$-
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Electricity Price ($ per MWH)
$140
$9
Natural Gas Price ($ per MMBtu)
$180
2009
2010
2011
2012
Energy, ancillary services and uplift costs are the three components in the all-in price of
electricity. The ERCOT wide price is the load weighted average of the real-time market prices
from all load zones. Prior to ERCOT’s conversion to the nodal market in December 2010,
energy costs were determined from the zonal balancing energy market. Ancillary services costs
are estimated based on total system demand and prices in the ERCOT markets for regulation,
responsive reserves, and non-spinning reserves. Uplift costs are assigned market-wide on a loadPage 1
Real-Time Market
ERCOT 2012 State of the Market Report
ratio share basis to pay for charges associated with additional reliability unit commitment and
any reliability must run contracts. 5
Figure 1 shows the monthly average all-in price for all of ERCOT from 2009 to 2012 and the
associated natural gas price. This figure indicates that natural gas prices were a primary driver of
the trends in electricity prices from 2009 to 2012. Again, this is not surprising given that natural
gas is a widely-used fuel for the production of electricity in ERCOT, especially among
generating units that most frequently set locational marginal prices in the nodal market.
The largest component of the all-in cost of wholesale electricity is the energy cost, which is
reflected by the locational marginal prices. As is typical in other wholesale electricity markets,
only a small share of the power produced in ERCOT is transacted in the spot market. However,
prices in the real-time energy market are very important because they set the expectations for
prices in the forward markets (including bilateral markets) where most transactions take place.
Unless there are barriers preventing arbitrage of the prices between the spot and forward
markets, the prices in the forward market should be directly related to the prices in the spot
market (i.e., the spot prices and forward prices should converge over the long-run). Hence,
artificially low prices in the real-time energy market will translate to artificially-low forward
prices. Likewise, price spikes in the real-time energy market will increase prices in the forward
markets. This section evaluates and summarizes electricity prices in the real-time market during
2012.
To summarize the price levels during the past four years, Figure 2 shows the monthly loadweighted average prices in the four geographic ERCOT load zones. These prices are calculated
by weighting the energy price for each interval and each zone by the total zonal load in that
interval. Since December 2010 these prices were determined by the nodal real-time energy
market. Prior prices were derived from the zonal balancing energy market. Load-weighted
average prices are the most representative of what loads are likely to pay, assuming that real-time
energy prices are, on average, generally consistent with bilateral contract prices.
5
Prior to December 2010 uplift costs included charges for out-of-merit energy and capacity, replacement reserve
services and any reliability must run contracts.
Page 2
ERCOT 2012 State of the Market Report
Figure 2: Average Real-Time Energy Market Prices
$180
Average Real-Time Electricity Price
($ per MWh)
2009
2010
2011
2012
$160
Electricity Price ($ per MWh)
$140
$120
$100
Real-Time Market
ERCOT
Houston
North
South
West
$34.03
$34.76
$32.28
$37.13
$27.18
$39.40
$39.98
$40.72
$40.56
$33.76
$53.23
$52.40
$54.24
$54.32
$46.87
$28.33
$27.04
$27.57
$27.86
$38.24
Natural Gas
($/MMBtu)
$3.74
$4.34
$3.94
$2.71
2012
2011
$80
$60
$40
$20
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
$0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
ERCOT average real-time market prices were 47 percent lower in 2012 than in 2011. The
ERCOT-wide load-weighted average price was $28.33 per MWh in 2012 compared to
$53.23 per MWh in 2011. The decrease in real-time energy prices was correlated with much
lower fuel prices in 2012. The steady decline in natural gas prices from June 2011 to April 2012
resulted in the 2012 average natural gas price of $2.71 per MMBtu, a 31 percent decrease
compared to $3.94 per MMBtu in 2011.
To provide additional perspective on the outcomes in the ERCOT market, our next analysis
compares the all-in price metrics for ERCOT and other electricity markets. The following figure
compares the all-in prices in ERCOT with other organized electricity markets in the U.S.: New
York ISO, ISO New England, PJM, Midwest ISO, and California ISO.
Page 3
Real-Time Market
ERCOT 2012 State of the Market Report
Figure 3: Comparison of All-in Prices across Markets
$70
Capacity
Uplift
Ancillary Services
Energy
Electricity Price ($ per MWh)
$60
$50
$40
$30
$20
$10
2009
2010
2011
2012
2009
2010
2011
2012
2009
2010
2011
2012
2009
2010
2011
2012
2009
2010
2011
2012
2009
2010
2011
2012
$-
ERCOT
NYISO
Average
ISO-NE Hub
PJM
Average
MISO
Average
CAISO
For each region, the figure reports the average cost (per MWh of load) for energy, ancillary
services (reserves and regulation), capacity markets (if applicable), and uplift for economically
out-of-merit resources. Figure 3 shows that ERCOT all-in prices in 2012 were roughly
equivalent to the Midwest ISO and significantly lower than all other regions. Although prices in
all markets declined from 2011 to 2012, no other region experienced anything close to the
magnitude of reduction seen in ERCOT.
Figure 4 presents price duration curves for ERCOT energy markets in each year from 2009 to
2012. A price duration curve indicates the number of hours (shown on the horizontal axis) that
the price is at or above a certain level (shown on the vertical axis). The prices in this figure are
the hourly load-weighted zonal balancing energy price for the zonal market and hourly loadweighted nodal settlement point price for the nodal market. 6
6
ERCOT switched to a nodal market on December 1, 2010. The December nodal prices are included in the 2010
price duration curve.
Page 4
ERCOT 2012 State of the Market Report
Real-Time Market
Figure 4: ERCOT Price Duration Curve
Due to the lowest natural gas prices seen in ten years, the 2012 price duration curve is below the
duration curve of other years in most hours.
To see where the prices during
Figure 5: ERCOT Price Duration Curve
Top 5% of Hours
2012 were much different than
$3,500
present Figure 5, which
$3,000
compares prices for the highest
five percent of hours. In 2011,
energy prices for the top 100
hours were significantly higher
due to higher loads leading to
more shortage conditions
coupled with a more effective
shortage pricing mechanism
Electricity price ($ per MWh)
in the previous three years, we
$2,500
2012
2011
2010
2009
$2,000
$1,500
$1,000
$500
$0
0
100
200
Hours
300
400
Page 5
Real-Time Market
ERCOT 2012 State of the Market Report
implemented as part of the nodal market design. In 2012, the energy duration curve for the top
five percent of hours is lower than the past three years for the majority of hours, reflecting lower
loads and resulting fewer occassions of shortage conditions. However, during the brief periods
of shortage that were experienced in 2012 prices rose to levels higher than those experienced in
2009 and 2010 when all prices in the zonal market were the result of actual submitted offers.
To better observe the effect of the highest-priced hours, the following analysis focuses on the
frequency of price spikes in the real-time energy market. Data prior to December 2010 is from
the zonal balancing energy market. Figure 6 shows the average price and the number of price
spikes in each month. For this analysis, price spikes are defined as intervals where the loadweighted average energy price in ERCOT is greater than 18 MMBtu per MWh times the
prevailing natural gas price. Prices at this level have historically exceeded the marginal costs of
virtually all of the on-line generators in ERCOT.
Figure 6: Average Real-Time Energy Prices and Number of Price Spikes
Page 6
ERCOT 2012 State of the Market Report
Real-Time Market
The number of price spike intervals during 2012 was 94 per month, an increase from the 83 per
month in 2011. However, just looking at the average can be misleading. Due to extreme
weather in February and August 2011, there were only two months with very high numbers of
price spikes in 2011. In contrast, all months from March to October 2012 had at least 85 price
spikes. As discussed later in this section, the high number of price spikes in 2012 is likely
related to the very low price of natural gas and resulting ‘overlap’ of offers from natural gas and
coal-fired units.
To measure the impact of these price spikes on average price levels, the figure also shows
average prices with and without the price spike intervals. The top portions of the stacked bars
show the impact of price spikes on monthly average price levels. Prior to 2012, the impact grew
with the frequency of the price spikes, averaging $4.67, $5.53 and $14.09 per MWh during 2009,
2010 and 2011, respectively. Although the frequency of price spikes increased in 2012, the
magnitude of their price impact decreased. The impact on average energy price in 2012 declined
to $3.63 per MWh, or 16 percent of the annual average price. This is explained by much lower
natural gas prices in 2012, resulting in a much lower threshold level for the definition of a “price
spike”.
To depict how real-time energy prices vary by hour in each zone, Figure 7 below shows the
hourly average price duration curve in 2012 for four ERCOT load zones. The Houston, North
and South load zones had similar prices over the majority of hours. The price duration curve for
the West zone is noticeably different than the other zones, with more hours with prices greater
than $50 per MWh and more than 500 hours (6 percent of the time) when the average hourly
price was less than zero. As observed over the past few years, West zone prices are lower than
the rest of ERCOT when high wind output in the West results in congested transmission
interfaces from the West zone to the other zones in ERCOT. Recently, prices higher than the
rest of ERCOT have occurred in the West zone due to local transmission constraints that
typically occur under low wind and high load, or outage conditions.
Page 7
Real-Time Market
ERCOT 2012 State of the Market Report
Figure 7: Zonal Price Duration Curves
$300
Houston
North
South
West
Electricity Price ($ per MWh)
$250
$200
< $0
3
6
3
571
$0-$50
8610
8573
8555
6741
Frequency of Prices
$50-$100
$100-$200
91
49
125
49
128
65
1103
191
> $200
31
31
33
178
$150
Houston
North
South
West
$100
$50
$0
($50)
0
1,000
2,000
3,000
4,000
5,000
Number of Hours
6,000
7,000
8,000
Figure 8 below shows the relationship between West zone and ERCOT average prices for the
2009 through 2012.
Figure 8: West Zone and ERCOT Price Duration Curves
$100
2012
$80
$20
$0
-$20
ERCOT
West
$60
$40
$20
$0
-$20
-$40
0
$100
1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
-$40
0
Number of Hours
2010
$80
ERCOT
$100
$20
$0
Number of Hours
ERCOT
West
$60
$ per MWh
$40
1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
2009
$80
West
$60
$ per MWh
2011
$80
$ per MWh
$40
$40
$20
$0
-$20
-$20
-$40
-$40
0
Page 8
$100
West
$60
$ per MWh
ERCOT
1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
Number of Hours
0
1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
Number of Hours
ERCOT 2012 State of the Market Report
Real-Time Market
On the low price end, we observe a reduction in the number of hours when West zone prices
were below the ERCOT average. We also note that minimum West zone prices have increased;
that is, become “less negative”. During 2012, for the first time, West zone prices were much
higher than ERCOT average for a significant number of hours. The combination of more hours
with higher prices, and fewer hours with less negative prices resulted in the average real-time
energy price in the West zone being greater than the ERCOT average.
More details about the transmission constraints influencing energy prices in the West zone are
provided in Section III, Transmission and Congestion.
B.
Real-Time Prices Adjusted for Fuel Price Changes
Although real-time electricity prices are driven to a large extent by changes in fuel prices, natural
gas prices in particular, they are also influenced by other factors.
Figure 9: Implied Marginal Heat Rate Duration Curve – All hours
To clearly identify changes in electricity prices that are not driven by changes in natural gas
prices, Figure 9 and Figure 10 show the load weighted, hourly average real-time energy price
adjusted to remove the effect of natural gas price fluctuations. The first chart shows a duration
Page 9
Real-Time Market
ERCOT 2012 State of the Market Report
curve where the real-time energy price is replaced by the marginal heat rate that would be
implied if natural gas was always on the margin. 7
The second chart shows the same duration curves for the five percent of hours in each year with
the highest implied heat rate. Both figures show duration curves for the implied marginal heat
rate for 2009 to 2012. In contrast to Figure 4 where the 2012 price duration curve lies below the
curves of other years, Figure 9 shows that the implied marginal heat rates were higher in 2012 as
compared to the three prior
years. This can be explained
Figure 10: Implied Marginal Heat Rate Duration Curve –
Top five percent of hours
by the much lower natural gas
Figure 10 shows the implied
marginal heat rates for the top
five percent of hours in 2009
through 2012 and highlights
that the implied heat rate in
2012 at the top 5 percent of
1,200
Implied Heat Rate (MMBtu / MWh)
prices in 2012.
1,000
hours is consistent with other
years, except for 2011, where
800
2012
2011
2010
2009
600
400
200
0
0
100
200
Hours
300
400
the heat rates were higher at top hours.
To further illustrate these differences, the next figure shows the implied marginal heat rates on a
monthly basis in each of the ERCOT zones in 2011 and 2012, with annual average heat rate data
for 2009 through 2012. This figure is the fuel price-adjusted version of Figure 2 in the prior subsection. Adjusting for natural gas price influence, Figure 11 shows that the annual, system-wide
average implied heat rate decreased in 2012 compared to 2011. However, it was still higher than
2009 and 2010.
7
The Implied Marginal Heat Rate equals either the Balancing Energy Price (zonal) or the Real-Time Energy
Price (nodal) divided by the Natural Gas Price. This methodology implicitly assumes that electricity prices
move in direct proportion to changes in natural gas prices.
Page 10
ERCOT 2012 State of the Market Report
Real-Time Market
Figure 11: Monthly Average Implied Heat Rates
45
2009
35
30
ERCOT
Houston
North
South
West
Natural Gas
2012
9.1
9.3
8.6
9.9
7.3
9.1
9.2
9.4
9.3
7.8
13.5
13.3
13.7
13.8
11.9
10.5
10.0
10.2
10.2
14.1
$3.74
$4.34
$3.94
$2.71
2012
2011
25
20
15
10
5
0
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Implied Marginal Heat Rate (MMBtu per MWh)
40
Average Heat Rates
2010
2011
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
The monthly average implied heat rates in 2012 are generally consistent with 2011, with notable
exceptions in February and August 2011. Higher heat rates in February can be explained by the
extended period when real-time prices were $3,000 per MWh due to extreme cold weather and
the resulting unplanned outages of numerous generators. Extended hot, dry weather resulted in
record system peak demands in August, and another extended period of energy prices reflecting
shortage conditions. The differences in the average annual implied heat rates observed at the
zonal level can be attributed to the continued significant congestion related to wind generation
exports from the West zone.
We conclude our examination of implied heat rates from the real-time energy market by
evaluating them at various load levels. Figure 12 below, provides the average heat rate at various
system load levels from 2010 through 2012. 8
8
To appropriately compare twelve months of data under each market design, data labeled as 2010 in Figure 12
are from December 1, 2009 through November 30, 2010.
Page 11
Real-Time Market
ERCOT 2012 State of the Market Report
Figure 12: Heat Rate and Load Relationship
In a well performing market, a clear positive relationship between these two variables is
expected since resources with higher marginal costs must be dispatched to serve higher loads.
Although we do see a generally positive relationship, there is a noticeable disparity for loads
between 50 and 55 GW. During the extreme cold weather event in early February 2011, loads
were at this level while prices reached $3,000 per MWh for a sustained period of time. Focusing
on 2012 data, we observe the desired positive relationship between load and implied heat rates.
The higher heat rates observed at lower loads in 2012 are likely due to the interplay between coal
and natural gas prices because of the low natural gas prices experienced in 2012. This
interaction warrants more explanation. The price of energy offered from coal units is generally
very stable, due to the long term nature of contracts for both fuel and transportation. The large
majority of energy offered from coal units is generally priced between $5 and $30 per MWh.
The price of energy from natural gas-fired units in ERCOT is much more variable but closely
tied to the price of natural gas fuel. In fact, the implied heat rate (the measure of conversion
efficiency from MMBtu of fuel to MWh of electricity) of natural gas based offers has remained
Page 12
ERCOT 2012 State of the Market Report
Real-Time Market
within a fixed range of 7.5 MMBtu per MWh to 18 MMBtu per MWh. Focusing on the past two
years, natural gas prices peaked in June 2011 at $4.54 per MMBtu and declined steadily to a
nadir of $1.88 per MMBtu in April 2012. At these low prices, energy offers from natural gas
units competed directly with
offers from coal units. Figure 13
Figure 13: Typical Energy Offers
90
energy offers from coal units with
80
those from natural gas units, under
70
both high and low natural gas
60
prices. As discussed later, one of
the effects of this price overlap
was reduced generation from coal
units during 2012. At natural gas
prices above $3 per MMBtu the
amount of overlap between energy
offers from coal and natural gas
$ per MWh
compares the typical ranges of
50
40
30
20
10
0
Coal
Low
Gas Price
High
Gas Price
units decreases.
C.
Real-Time Price Volatility
Volatility in real-time wholesale electricity markets is expected because system load can change
rapidly and the ability for supply to adjust can be restricted by physical limitations of the
resources and the transmission network. Figure 14 below presents a view of the price volatility
experienced in ERCOT’s real-time energy market during the summer months of May through
August. Average five-minute real-time energy prices are presented along with the magnitude of
change in price for each five-minute interval. Average real-time energy prices from the same
period in 2011 are also presented. Comparing average real-time energy prices for 2012 with
those from 2011, the effects of lower natural gas prices on average prices during non-peak hours
and the effects of fewer shortage intervals during peak hours are observed. Outside of the hours
from 15 to 18 (2:00 pm to 6:00 pm), short-term increases in average real-time energy prices are
typically due to high prices resulting from generator ramp rate limitations occurring at times
when significant amounts of generation is changing its online status. Factoring current market
conditions into generators’ daily operational decisions about the specific timing of startup and
Page 13
Real-Time Market
ERCOT 2012 State of the Market Report
shutdown may have led to the reduced the effects of this type of ramp rate limitation on price
spikes during 2012. The average of the absolute value of changes in five-minute real-time
energy prices, expressed as a percentage of average price was less than 4 percent in 2012
compared to approximately 6 percent for the same period in 2011.
Figure 14: Real-Time Energy Price Volatility (May – August)
$350
$140
$300
$120
Average LMP Change - 2012
Average LMP - 2012
$100
Average LMP - 2011
$200
$80
$150
$60
$100
$40
$50
$20
$0
$0
-$50
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Change ($ per MWh)
$ per MWh
$250
-$20
Reduced price volatility in 2012 is also observed in 15 minute settlement point prices for the four
geographic load zones, as shown below in Table 1.
Table 1: 15 Minute Price Changes as a Percent of Annual Average Price
Load Zone
Houston
South
North
West
2011
21.4%
19.9
22.5
26.2
2012
13.0%
13.1
13.9
19.4
The table shows that the price volatility fell substantially from 2011 to 2012. This was primarily
due to the sharply reduction in shortages that occurred in 2012, which exhibit relatively normal
Page 14
ERCOT 2012 State of the Market Report
Real-Time Market
summer weather conditions. In contrast, 2011 exhibited the hottest summer temperatures in
more than 100 years, leading to frequent shortages and associated higher price volatility. The
table also shows that price volatility in the West zone has continued to be higher than in the other
zones, which is expected given the very high penetration of variable output wind generation
located in that area.
D.
Prices at the System-Wide Offer Cap
After the extremes of 2011, weather conditions in Texas returned to closer to normal in 2012.
As more fully discussed in Section IV Load and Generation, overall demand for electricity was
lower in 2012 than in 2011, resulting in much fewer occasions when the available supply
generation capacity was unable to meet customer demands. This resulted in a decreased
likelihood that the available generation capacity was not sufficient to meet customer demands for
electricity and maintain the required reliability reserves.
As more fully described later in Section V, Resource Adequacy, the nodal market causes energy
prices to rise toward the system-wide offer cap as available operating reserves approach
minimum required levels to reflect the degradation in system reliability.
Figure 15: Duration of Prices at the System-Wide Offer Cap
20
17.4
15
10
6.0
5
2011
Dec
Nov
Oct
Sep
Aug
0.1
Jul
Jun
May
0.4
Apr
Mar
Feb
Jan
Dec
Oct
Sep
1.0
0.7
0.2
Nov
0.2
Aug
1.1
Jul
0.6
Jun
Mar
Feb
0
0.1
May
0.6
Apr
1.5
Jan
Hours
Prices were at the System Wide Offer Cap for
28.4 hours in 2011 and
1.5 hours in 2012
2012
Page 15
Real-Time Market
ERCOT 2012 State of the Market Report
Figure 15 above shows the aggregated amount of time represented by all five-minute dispatch
intervals where the real-time energy price was at the system-wide offer cap, displayed by month.
Prices during 2012 prices were at the system-wide offer cap for only 1.5 hours, a significant
reduction from the 28.4 hours experienced in 2011. Approved during 2012, PUCT SUBST.
R.25.508 increased the system-wide offer cap to $4,500 per MWh effective August 1, 2012. As
shown in Figure 15 above, there was only a brief period when energy prices rose to the cap after
this change was implemented
The next figure provides a detailed comparison of August’s load, required reserve levels, and
prices for 2011 and 2012. As expected, the weather in ERCOT was extremely hot and dry
during both months, but there were very few dispatch intervals when real-time energy prices
reached the system-wide offer cap in 2012 compared to the relatively high frequency it occurred
in 2011. Although the weather may have been similar, there were significant differences in load
and available operating reserve levels, resulting in much higher prices in August 2011.
Figure 16: Load, Reserves and Prices: August 2011 and August 2012
Page 16
ERCOT 2012 State of the Market Report
Real-Time Market
Shown on the left side of Figure 16 is the relationship between real-time energy price and load
level for each dispatch interval for the month of August in 2011 (top) and 2012 (bottom).
ERCOT loads were greater than 65 GW for 71 hours in 2011, whereas loads reached that level
for less than 12 hours during August 2012. As previously discussed, a strong positive correlation
between higher load and higher prices is expected in a well-functioning energy market. We
observe such a relationship between higher prices and higher loads in both months.
Although load levels are strong predictors of energy prices, an even more important predictor is
the level of operating reserves. Simply put, operating reserves are the difference between the
total capacity of operating resources and the current load level. As load level increases against a
fixed quantity of operating capacity, the amount of operating reserves diminishes. The minimum
required operating reserves prior to the declaration of Energy Emergency Alert (“EEA”) Level 1
by ERCOT is 2,300 MW. As the available operating reserves approach the minimum required
amount, energy prices should rise toward the system-wide offer cap to reflect the degradation in
system reliability.
On the right side of Figure 16 are data showing the relationship between real-time energy prices
and the quantity of available operating reserves for each dispatch interval during August of 2011
(top) and 2012 (bottom). This figure shows a strong correlation between diminishing operating
reserves and rising prices. With the lower loads in August 2012, available operating reserves
were well above minimum levels for the entire month, and there were no occurrences where the
energy price reached the system-wide offer cap. In contrast, there were numerous dispatch
intervals in August 2011 when the minimum operating reserve level was approached or
breached, with 17.4 hours where prices reached $3,000 per MWh. 9 It should be noted that
during August 2011 there were a number of dispatch intervals where operating reserves were
below minimum requirements with prices well below the level that would be reflective of the
reduced state of reliability. In Section IV, Load and Generation at page 93, we provide an
example explaining why this can occur and offer a recommendation for improvement.
9
The system-wide offer cap during August 2011 was set at $3,000 per MWh. It was increased to $4,500 per MWh
effective August 1, 2012.
Page 17
Real-Time Market
E.
ERCOT 2012 State of the Market Report
Mitigation
The dispatch software includes an automatic, two step price mitigation process. In the first step,
the dispatch software calculates output levels (Base Points) and associated locational marginal
prices using the participants’ offer curves and only considering transmission constraints that have
been deemed competitive. These “reference prices” at each generator location are compared
with that generator’s mitigated offer cap, and the higher of the two is used to formulate the offer
curve to be used for that generator in the second step in the dispatch process. The resulting
mitigated offer curve is used by the dispatch software to determine the final output levels for
each generator taking all transmission constraints into consideration. This approach is intended
to limit the ability of a specific generator to raise prices in the event of a transmission constraint
that requires their output to resolve. In this section we analyze the quantity of capacity affected
by this mitigation process.
Our first analysis computes how much capacity, on average, is actually mitigated during each
dispatch interval. The results, shown in Figure 17, are provided by load level.
Figure 17: Mitigated Capacity by Load Level
400
0.60%
Mitigated MW 2011
2011 Mitigated MW as % Dispatched Capacity
0.50%
2012 Mitigated MW as % Dispatched Capacity
300
0.40%
MW
250
200
0.30%
150
0.20%
100
0.10%
50
0
0.00%
<25
Page 18
25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 60 - 65
Load Level (GW)
>65
Percentage of Base Points being Mitigated
Mitigated MW 2012
350
ERCOT 2012 State of the Market Report
Real-Time Market
The quantities of capacity actually mitigated in 2012 were much larger than during 2011,
averaging 14 MW at low loads and increasing to 338 MW at loads above 65 GW. Although the
quantities of mitigated capacity were greater in 2012, they were less than one-half of one percent
of the total dispatched capacity at all but the very highest load levels. The decrease in mitigated
capacity at high loads observed in 2011 due to the reluctance by ERCOT operators to activate
certain transmission constraints during very high system load conditions is not present in 2012.
In the previous figure only the amount of capacity that can be dispatched within one interval is
counted as mitigated. In our next analysis we compute the total capacity subject to mitigation.
These values are determined by comparing a generator’s mitigated and unmitigated (as
submitted) offer curves and determining the point at which they diverge. We then take the
difference between the total unit capacity and the capacity at the point the curves diverge. This
calculation is performed for all units and aggregated by load level, as shown in Figure 18. From
this figure we observe that at most 7 percent of capacity necessary to serve load is subject to
mitigation. An important note about this capacity measure is that it includes all capacity above
the point at which a unit’s offers become mitigated, without regard for whether that capacity is
actually required to serve load.
Figure 18: Capacity Subject to Mitigation
3,500
2011
2012
Capacity Subject to Mitigation (MW)
3,000
2,500
2,000
1,500
1,000
500
0
<25
25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 60 - 65
Load Level (GW)
>65
Page 19
Real-Time Market
ERCOT 2012 State of the Market Report
Although executing all the time, the automatic price mitigation aspect of the two step dispatch
process only has an effect when a non-competitive transmission constraint is active. This can
result in mitigating certain units inappropriately. The mitigation process is intended to limit the
ability of a generator to affect price when their output is required to manage congestion. The
process does not currently identify a situation where there are a competitively sufficient number
of generators on the other side of the constraint and mitigates all their offers. This unnecessary
mitigation will be addressed with the implementation of changes described in NPRR520. This
change will introduce an impact test to determine whether units are relieving or contributing to a
transmission constraint, and only subject the relieving units to mitigation.
Page 20
ERCOT 2012 State of the Market Report
II.
Day-Ahead Market
REVIEW OF DAY-AHEAD MARKET OUTCOMES
ERCOT’s centralized day-ahead market allows participants to make financially binding forward
purchases and sales of power for delivery in real-time. Offers to sell can take the form of either a
three-part supply offer, which allow sellers to reflect the unique financial and operational
characteristics of a specific generation resource, or an energy only offer, which is location
specific but is not associated with a generation resource. Bids to buy are also location specific.
Ancillary services are also procured as part of the day-ahead market clearing. The third type of
transaction included in the day-ahead market is bids to buy Point to Point (“PTP”) Obligations,
which allow parties to hedge the incremental cost of congestion between day-ahead and real-time
operations.
With the exception of the acquisition of ancillary service capacity, the day-ahead market is a
financial market. Although all bids and offers are evaluated in the context of their ability to
reliably flow on the transmission network, there are no operational obligations resulting from the
day-ahead market clearing. These transactions are made for a variety of reasons, including
satisfying the participant’s own supply, managing risk by hedging the participant’s exposure to
the real-time market, or arbitraging with the real-time markets. For example, load serving
entities can insure against volatility in the real-time market by purchasing in the day-ahead
market. Finally, the day-ahead market plays a critical role in coordinating generator
commitments. For all of these reasons, the performance of the day-ahead market is essential.
In this section we review energy pricing outcomes from the day-ahead market and compare their
convergence with real-time energy prices. We will also review the volume of activity in the dayahead market, including a discussion of PTP Obligations. We conclude this section with a
review of the ancillary service markets.
A.
Day-Ahead Market Prices
One indicator of market performance is the extent to which forward and real-time spot prices
converge over time. Forward prices will converge with real-time prices when two main
conditions are in place: a) there are low barriers to shifting purchases and sales between the
forward and real-time markets; and b) sufficient information is available to market participants to
Page 21
Day-Ahead Market
ERCOT 2012 State of the Market Report
allow them to develop accurate expectations of future real-time prices. When these conditions
are met, market participants can be expected to arbitrage predictable differences between
forward prices and real-time spot prices by increasing net purchases in the lower-priced market
and increasing net sales in the higher-priced market. These actions will tend to improve the
convergence of forward and real-time prices.
In this section, we evaluate the price convergence between the day-ahead and real-time markets.
This analysis reveals whether persistent and predictable differences exist between day-ahead and
real-time prices, which participants should arbitrage over the long-term. To measure the shortterm deviations between real-time and day-ahead prices, we also calculate the average of the
absolute value of the difference between the day-ahead and real-time price on a daily basis.
This measure captures the volatility of the daily price differences, which may be large even if the
day-ahead and real-time energy prices are the same on average. For instance, if day-ahead prices
are $70 per MWh on two consecutive days while real-time prices are $40 per MWh and $100 per
MWh on the two days, the absolute price difference between the day-ahead market and the realtime market would be $30 per MWh on both days, while the difference in average prices would
be $0 per MWh.
Figure 19 shows the price convergence between the day-ahead and real-time market,
summarized by month. Day-ahead prices averaged $29 per MWh in 2012 compared to an
average of $27 per MWh for real-time prices. 10 The average absolute difference between dayahead and real-time prices was $9.96 per MWh in 2012; much lower than in 2011 when average
of the absolute difference was $24.50 per MWh. This reduction was due to fewer occurrences of
shortage intervals and associated high prices in 2012. This day-ahead premium is consistent
with expectations due to the much higher volatility of real-time prices. Risk is lower for loads
purchasing in the day-ahead and higher for generators selling day-ahead. The higher risk for
generators is associated with the potential of incurring a forced outage and as a result, having to
buy back energy at real-time prices. This explains why the highest premiums tend to occur
during the months with the highest demand and highest prices. Overall, the day-ahead premiums
10
These values are simple averages, rather than load-weighted averages presented in Figure 1and Figure 2.
Page 22
ERCOT 2012 State of the Market Report
Day-Ahead Market
were very similar to the differences observed in 2010 and 2011, but remain higher than observed
in other organized electricity markets. 11 Real-time energy prices in ERCOT are allowed to rise
to levels that are much higher than what is allowed in other organized electricity markets, which
increases risk and helps to explain the higher day ahead premiums regularly observed in ERCOT
Although most months experienced a day-ahead premium (e.g., $10 per MWh in August), it
should not be expected over time that every month will always produce a day-ahead premium as
the real-time risks that lead to the premiums will materialize unexpectedly on occasion, resulting
in real-time prices that exceed day-ahead prices (e.g., in March).
Figure 19: Convergence between Forward and Real-Time Energy Prices
60
50
$ per MWh
40
2009
2010
2011
2012
Average
Average
Day-Ahead Real Time
Price
Price
$38
$35
$42
$40
$46
$43
$29
$27
Day Ahead
Real Time
Absolute Difference
30
20
10
0
Jan
11
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
In 2009 and 2010 under the zonal market the comparison was made between on-peak forward prices and prices
for the same on-peak period in the balancing energy market.
Page 23
Day-Ahead Market
ERCOT 2012 State of the Market Report
In Figure 20 below, monthly day-ahead and real-time prices are shown for each of the
geographic load zones. Of note is the difference in the West zone data compared to the other
regions. The higher volatility in West zone pricing is likely associated with the uncertainty of
forecasting wind generation output and the resulting price differences between day-ahead and
real-time.
Figure 20: Day-Ahead and Real-Time Prices by Zone
80
Average
Day-Ahead
Price
Houston
28
28
North
28
South
37
West
70
60
$ per MWh
50
Average
Real-Time
Price
25
25
26
35
Day Ahead
Absolute
Difference
8
8
9
20
Real Time
Absolute Difference
40
30
20
10
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
0
Jan
B.
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Day-Ahead Market Volumes
Our next analysis summarizes the volume of day-ahead market activity by month. In Figure 21
below, we find that day-ahead purchases are approximately 45 percent of real-time load. These
energy purchases are met through a combination of generator specific and virtual offers.
As discussed in more detail in the next sub-section, Point to Point Obligations are financial
instruments purchased in the day-ahead market. They do not provide any energy supply
themselves, but they do provide the ability to avoid the congestion costs associated with
Page 24
ERCOT 2012 State of the Market Report
Day-Ahead Market
transferring the delivery of energy from one location to another. To provide a volume
comparison we aggregate all of these “transfers”, netting location specific injections against
withdrawals. By adding the aggregated transfer capacity associated with purchases of PTP
Obligations, we find that on average, total volumes transacted in the day-ahead market are
greater than real-time load.
Figure 21: Volume of Day-Ahead Market Activity by Month
70
Day Ahead Market Volume (GW)
60
Three Part Awards
Energy Only Awards
Load
PTP Obligations
Day-Ahead Purchase
50
40
30
20
10
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Figure 22 below, presents the same day-ahead market activity data summarized by hour of the
day. In this figure the volume of day-ahead market transactions is disproportionate with load
levels between the hours of 6 and 22. Since these times align with common bilateral transaction
terms, it appears that market participants are using the day-ahead market to trade around those
positions.
Page 25
Day-Ahead Market
ERCOT 2012 State of the Market Report
Figure 22: Volume of Day-Ahead Market Activity by Hour
60
Three Part Awards
Energy Only Awards
Load
PTP Obligations
Day-Ahead Purchase
Day Ahead Market Volume (GW)
50
40
30
20
10
0
1
C.
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour of the Day
Point to Point Obligations
Purchases of Point to Point (“PTP) Obligations comprise a significant portion of day-ahead
market activity. These instruments are similar to, and can be used to complement Congestion
Revenue Rights (CRRs). CRRs, as more fully described in Section III, are acquired via monthly
and annual auctions and allocations. CRRs accrue value to their owner based on locational price
differences as determined by the day-ahead market. PTP Obligations are instruments that are
purchased as part of the day-ahead market and accrue value to their owner based on real-time
locational price differences. By acquiring a PTP Obligation using the proceeds due to them from
their CRR holding, a market participant may be described as rolling their hedge to real-time.
In this subsection we provide additional details about the volume and profitability of these PTP
Obligations.
Page 26
ERCOT 2012 State of the Market Report
Day-Ahead Market
Figure 23: Point to Point Obligation Volume
70
Generation Hedge
Physical Parties
Financial Parties
60
Total PTP volume (GW)
50
40
30
20
10
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Figure 23 presents the total volume of PTP Obligation purchases divided into three categories.
Compared to the previous two figures which showed net flows associated with PTP Obligations,
in this figure we examine the total volume. For all PTP Obligations that source at a generator
location, we attribute capacity up to the actual generator output as a generator hedge. From the
figure above we see that this comprised most of the volume of PTP Obligations purchased. The
remaining volumes of PTP Obligations are not directly linked to a physical position and are
assumed to be purchased primarily to profit from anticipated price differences between two
locations. We further separate this arbitrage activity by type of market participant. Physical
parties are those that have actual real-time load or generation, whereas financial parties have
neither.
To the extent the price difference between the source and sink of a PTP Obligation is greater in
real-time than it was for the day-ahead price, the owner will have made a profit. Conversely, if
the price difference does not materialize in real-time, the PTP Obligation may be unprofitable.
Page 27
Day-Ahead Market
ERCOT 2012 State of the Market Report
We compare the profitability of PTP Obligation holdings by the two types of participants in
Figure 24.
Figure 24: Average Profitability of Point to Point Obligations
0.6
Physical Parties
Financial Parties
Average Profit ($ per MW-Hour)
0.5
0.4
0.3
0.2
0.1
0.0
(0.1)
(0.2)
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
From the figure above we can infer different motivations between the two types of participants.
Because financial participants have no real-time load or generation they have no other exposure
to real-time prices. If a financial participant is not making a profit on their PTP Obligations there
is no reason for them to buy any. In fact, their profit seeking action of buying PTP Obligations
between points where congestion is expected helps make the day-ahead market converge with
real-time market outcomes. On the other hand, physical participants do have exposure to realtime prices. It is reasonable to expect that this type of participant is most interested in limiting
that exposure by using PTP Obligations as a hedge.
Page 28
ERCOT 2012 State of the Market Report
Day-Ahead Market
Figure 25: Point to Point Obligation Charges and Payments
100
Day Ahead Charge
Real-Time Payment
90
2012 Annual Charge: $405.8 Million
2012 Annual Payment: $487.0 Million
80
2011 Annual Charge: $296.5 Million
2011 Annual Payment: $460.8 Million
70
Million $
60
50
40
30
20
10
2011
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
0
2012
To conclude our analysis of PTP Obligations, in Figure 25 we compare the total amount paid for
these instruments day-ahead, with the total amount received by their holders in real-time. In
most months owners received, in aggregate, more in payments for their PTP Obligations than
they paid to acquire them. This occurs when real-time congestion is greater than what occurred
in the day-ahead. The payments made to PTP Obligation owners come from real-time
congestion rent. We assess the sufficiency of real-time congestion rent to cover both PTP
Obligations and payments to owners of CRRs who elect to receive payment based on real-time
prices in Section III, Transmission and Congestion at page 52.
D.
Ancillary Services Market
The primary ancillary services are up regulation, down regulation, responsive reserves, and nonspinning reserves. Market participants may self-schedule ancillary services or purchase their
required ancillary services through the ERCOT markets. This section reviews the results of the
Page 29
Day-Ahead Market
ERCOT 2012 State of the Market Report
ancillary services markets in 2012. We start with a display of the quantities of each ancillary
service procured each month shown in Figure 26.
6,000
Figure 26: Ancillary Service Capacity
Regulation Down
Regulation Up
Nonspin Reserve
Responsive Reserve
5,000
MW
4,000
3,000
2,000
1,000
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
In general, the purpose of responsive and non-spinning reserves is to protect the system against
unforeseen contingencies (e.g., unplanned generator outages, load forecast error, wind forecast
error), rather than for meeting normal load fluctuations. ERCOT procures responsive reserves to
ensure that the system frequency can quickly be restored to appropriate levels after a sudden,
unplanned outage of generation capacity. Non-spinning reserves are provided from slower
responding generation capacity, and can be deployed alone, or to restore responsive reserve
capacity. Regulation reserves are capacity that responds every four seconds, either increasing or
decreasing as necessary to fill the gap between energy deployments and actual system load.
One notable change made to ancillary service procurements during 2012 was the increased
amount of responsive reserve procured beginning in April. This 500 MW increase was balanced
with the same amount of decrease in the amount of non-spinning reserves procured. Although
Page 30
ERCOT 2012 State of the Market Report
Day-Ahead Market
the minimum level of required responsive reserve remains at 2,300 MW, having the additional
500 MW of responsive reserve provides a higher quality – that is, faster responding capacity
available to react to sudden changes in system conditions. As the amount of wind generation in
ERCOT continue to increase, this additional responsive reserve should contribute to improved
system reliability.
Under the nodal market, ancillary service offers are co-optimized as part of the day-ahead market
clearing. This means that market participants no longer have to include their expectations of
forgone energy sales in their ancillary services capacity offers. However, because clearing prices
for ancillary services capacity will explicitly account for the value of energy, there is a much
higher correlation between ancillary services prices and real-time energy prices.
Figure 27: Ancillary Service Prices
$25
$50
Nonspin Reserve
Energy Price
Regulation Up
$20
$40
$15
$30
$10
$20
$5
$10
$0
$0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Real Time Energy Price ($ per MWh)
Ancillary Service Price ($ per MWh)
Responsive Reserve
Regulation Down
Dec
With average energy prices varying only between $20 and $35 per MWh, we observe the prices
of ancillary services remaining fairly stable throughout the year. However, the average price of
responsive reserve was higher in April than in March, even though average energy price declined
Page 31
Day-Ahead Market
ERCOT 2012 State of the Market Report
for the same period. We attribute this pricing outcome to the increased responsive reserve
requirement being implemented in April.
In contrast to the previous data that showed the individual ancillary service capacity prices,
Figure 28 shows the monthly total ancillary service costs per MWh of ERCOT load and the
average real-time energy price for 2009 through 2012. This figure shows that total ancillary
service costs are generally correlated with real-time energy price movements, which, as
previously discussed, are highly correlated with natural gas price movements.
Average Price of 2009 2010 2011 2012
Ancillary Services $1.15 $1.26 $2.41 $1.06
$144
$8
$7
$6
$160
$128
$112
Real Time Electricity Price
Ancillary Service Prices
$96
$5
$80
$4
$64
$3
$48
$2
$32
$1
$16
$0
$0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Ancillary Service Price ($ per MWh of Load)
$9
Figure 28: Ancillary Service Costs per MWh of Load
Real Time Electricity Price ($ per MWh)
$10
2009
2010
2011
2012
The average ancillary service cost per MWh of load decreased to $1.06 per MWh in 2012
compared to $2.41 per MWh in 2011, a decrease of 56 percent. Total ancillary service costs
decreased from 4.5 percent of the load-weighted average energy price in 2011 to 3.7 percent in
2012.
Ancillary Service capacity is procured as part of the day-ahead market clearing. Between the
time it is procured and the time that it is required to be provided events can occur which make
Page 32
ERCOT 2012 State of the Market Report
Day-Ahead Market
this capacity unavailable to the system. Transmission constraints can arise which make the
capacity undeliverable. Outages or limitations at a generating unit can lead to failures to
provide. When either of these situations occurs, ERCOT may open a supplemental ancillary
services market (“SASM”) to procure replacement capacity. 12
Figure 29 presents a summary of the frequency with which ancillary service capacity was not
able to be provided and the number of times that a SASM was opened.
Figure 29: Frequency of SASM Clearing
100%
90%
80%
Percent of Hours
70%
60%
50%
40%
30%
20%
10%
A/S Normal
No SASM - A/S Failed or Infeasible
SASM - A/S Failed or Infeasible
2011
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
0%
2012
The percent of time that capacity procured in the day-ahead was actually able to provide the
service in the hour it was procured for continued to increase in 2012. This percentage was
52 percent in 2012 compared to 43 percent in 2011.
Even though in more than 40 percent of
the hours there were deficiencies in ancillary service deliveries, SASMs were opened to procure
replacement capacity only 7 percent of the time, down from 9 percent of the hours in 2011.
12
ERCOT may also open a SASM if they change their ancillary service plan. This did not occur during 2012.
Page 33
Day-Ahead Market
ERCOT 2012 State of the Market Report
In Table 2 below, we provide an annual summary of the frequency and quantity of ancillary
service deficiency, which is defined as either failure-to-provide or as undeliverable.
Table 2: Ancillary Service Deficiency
Hours
Deficient
Mean
Deficiency
(MW)
Median
Deficiency
(MW)
Responsive Reserve
3756
34
15
Non-Spin Reserve
664
36
8
Up Regulation
750
41
25
Down Regulation
522
48
39
Responsive Reserve
4053
39
20
Non-Spin Reserve
1254
90
39
Up Regulation
1222
27
20
Down Regulation
1235
22
11
2012
Service
2011
The number of hours with deficiency decreased for all types of ancillary services when compared
to 2011. Responsive Reserve service was deficient most frequently. Well over 90 percent of the
deficiency occurrences were caused by failure to provide by the resource rather than
undeliverability related to a transmission constraint. We also note that the average magnitude of
non-spin reserve deficiency reduced from 90 MW in 2011 to 36 MW in 2012. On the other
hand, the amount of deficiency in regulation services (up and down) increased slightly in 2012
when compared to 2011.
Page 34
ERCOT 2012 State of the Market Report
Day-Ahead Market
Figure 30: Ancillary Service Quantities Procured in SASM
400
Regulation Down
350
Regulation Up
Nonspin Reserve
300
Responsive Reserve
250
MW
200
150
100
50
2011
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
0
2012
Our final analysis in this section, shown in Figure 30, summarizes the average quantity of each
service that was procured via SASM. Along with reduced occurrences of ancillary service
deficiency, the quantity of services procured via SASM also declined in 2012. The average
quantities of responsive and non-spin reserves procured via SASM in 2012 were roughly half
what they were the previous year.
Page 35
Day-Ahead Market
Page 36
ERCOT 2012 State of the Market Report
ERCOT 2012 State of the Market Report
III.
Transmission and Congestion
TRANSMISSION AND CONGESTION
One of the most important functions of any electricity market is to manage the flows of power
over the transmission network by limiting additional power flows over transmission facilities
when they reach their operating limits. The action taken to ensure operating limits are not
violated is called congestion management. The effect of congestion management is to change
generator(s) output level so as to reduce the amount of electricity flowing on the transmission
line nearing its operating limit. Because the transmission system is operated such that it can
withstand the unexpected outage of any element at any time, congestion management actions
most often occur when a transmission element is expected to be overloaded if a particular
unexpected outage (contingency) were to occur. Congestion leads to higher costs due to lower
cost generation being reduced and higher priced generation increased. Different prices at
different nodes are the result. The decision about which generator(s) will vary their output is
based on generating unit specific offer curves and the relative shift factors to the contingency and
constraint pair. This leads to a dispatch of the most efficient resources available to reliably serve
demand.
After summarizing congestion activity in 2012, this section of the report provides a review of the
costs and frequency of transmission congestion in both the day-ahead and real-time markets. We
will then provide a review the activity in the congestion rights market.
A.
Summary of Congestion
There was a marked change in the nature of real-time transmission congestion during 2012 when
compared to previous years. For the past several years a significant portion of real-time
transmission congestion could be described as limiting the export of wind generation from the
West zone to the load centers across the rest of ERCOT. Transmission congestion in 2012 was
more significantly the result of limitations on the ability to get generation to loads in the West
zone. Some portion of the limitation can be attributed to transmission outages taken to enable
the construction of new CREZ transmission lines. Another factor is that loads in far west Texas
have experienced much greater growth than the system-wide average.
Page 37
Transmission and Congestion
ERCOT 2012 State of the Market Report
The total congestion revenue generated by the ERCOT real-time market in 2012 was
$480 million, a decrease of 9 percent from 2011. This decrease was not as large as might be
expected given the much larger decreases in average natural gas prices real-time energy prices.
Two factors influencing the overall costs of congestion in 2012 were the significant financial
impact of several localized transmission constraints in far west Texas and the higher frequency
of active transmission constraints.
Figure 31 provides a comparison of the amount of time transmission constraints were active at
various load levels in 2012 and 2011. Active transmission constraints are those for which
generators are being dispatched to a less efficient output level in order to maintain transmission
flows at reliable levels.
Figure 31: Frequency of Active Constraints
100%
90%
2011
2012
80%
Percent of Time
70%
60%
50%
40%
30%
20%
10%
0%
<25
25-30 30-35 35-40 40-45 45-50 50-55 55-60 60-65
Load (GW)
>65
Annual
We observe that in 2012 the likelihood of having an active transmission constraint was higher
than it was in 2011 and that for loads above 45 GW the frequency was much higher. During
2011, we observed that at higher system loads ERCOT operators did not always activate (or
sometimes de-activated) transmission constraints. This was due to a concern that by having a
constraint active during periods of high demand the total capacity available to serve load may be
Page 38
ERCOT 2012 State of the Market Report
Transmission and Congestion
limited. However, ERCOT’s dispatch software contains parameters that allow it to automatically
make the correct decision about when to violate transmission constraints when necessary to serve
total system load. Therefore, ERCOT modified their practice in 2012 to retain active
transmission constraints even during periods of high demand.
B.
Real-Time Constraints
We begin our review by examining the real-time transmission constraints with the highest
financial impact as measured by congestion rent. There were more than 350 different constraints
active at some point during 2012, a slight increase from 2011. The median financial impact, as
measured by congestion rent, was approximately $200,000 during 2012. This is a significant
decrease from 2011, when the median impact was approximately $300,000. Given the
significant reduction in average energy costs from 2011 to 2012, reduced financial impact of
congestion is expected.
Figure 32 below displays the ten most highly valued real-time constraints as measured by
congestion rent and indicates that the Odessa North 138/69 kV transformer constraint was by far
Figure 32: Top Ten Real-Time Constraints
Page 39
Transmission and Congestion
ERCOT 2012 State of the Market Report
the most highly valued during 2012. This constraint became more pronounced from 2011 to
2012 and is mainly attributed to load growth in far west Texas driven by increased oil and
natural gas activity.
The Odessa North 138/69 kV transformer typically overloads under low wind conditions. The
characteristics that load the Odessa North 138/69 kV transformer are the same conditions that
also affect the Odessa to Odessa North 138 kV line, which is shown as the seventh constraint on
the list. Not only did this constraint have nearly twice the financial impact of the second
constraint on the list, its impact was more than 40 percent greater than the top constraint from
2011. Its magnitude is even more significant given the overall lower costs of energy in 2012.
Not surprisingly, much public attention was focused on this constraint; much of it questioning its
causes and the potential for short-term remedies. ERCOT and the local transmission provider
were able to identify two transmission lines, which when opened greatly reduced congestion
around the Odessa North station without causing other reliability concerns. After the lines were
opened in mid-September, congestion around Odessa North was almost eliminated for the rest of
the year.
The second constraint on the list is the overload of the 138 kV transmission line between China
Grove and Bluff Creek. Also located in west Texas, west of Abilene, this constraint was new to
the top ten list in 2012. Most of the time this line was constrained in the China Grove to Bluff
Creek direction, typically under low wind conditions but also when there were other lines out of
service. Depending on the level of wind and outages, the direction of the constraint would be
reversed. That is, under high wind conditions flow on the line would be limited in the Bluff
Creek to China Grove direction (from the west).
The West to North constraint continued to be active at some point during every month of 2012
but with far less impact than in 2011. Through the years this constraint has been a major
impediment to delivering all the wind generation located and produced in the western reaches of
ERCOT to the load centers. This constraint had the highest financial impact of all real-time
transmission constraints during 2011, but in 2012 its impact dropped to one third that level. The
reduction was a result of the combined impact of higher loads in the West and increased transfer
Page 40
ERCOT 2012 State of the Market Report
Transmission and Congestion
capability due to the first CREZ transmission lines being placed in service. We will review the
utilization of this constraint in more detail later in this section.
In all, eight of the top ten real-time constraints during 2012 were due to load and wind conditions
within the West. In addition to the four constraints previously discussed, the Morgan Creek and
San Angelo Red Creek transformer constraints were typically active under high load and low
wind conditions. Unlike the previous constraints the Moore to Downie 138 kV line and the
Turnersville to Buda 138 kV line are not located in the West, but became constrained during low
wind and high load conditions in the West that occurred while other, larger capacity transmission
lines were out of service.
The remaining two constraints on the list were of fairly short duration; their high impact
reflecting limitations on the ability to import power to a major load center. The Lewisville to
Jones Street 138 kV constraint was related to serving load in the DFW area. Belton to Belton
Southwest was due to outage conditions that
limited the flow to Central Texas.
Figure 33: Real-Time Congestion Costs
ERCOT
4.8%
To further highlight the significance of West zone
HOUSTON
4.8%
congestion, Figure 33 depicts that more than half
NORTH
15.7%
of the costs of real-time congestion that occurred
in 2012 were associated with facilities located in
the West zone. Further, the cost of congestion on
facilities that spanned zonal boundaries, labeled
ERCOT in the figure, primarily comprised costs
WEST
55.8%
SOUTH
18.8%
associated with West to North congestion.
Irresolvable Constraints
When a constraint becomes irresolvable, ERCOT’s dispatch software can find no combination of
generators to dispatch in a manner such that the flows on the transmission line(s) of concern are
below where needed to operate reliably. In these situations, offers from generators are not
setting locational prices. Prices are set based on predefined rules, which since there are no
supply options for clearing, should reflect the value of reduced reliability for demand. To
Page 41
Transmission and Congestion
ERCOT 2012 State of the Market Report
address the situation more generally, a regional peaker net margin mechanism was introduced
such that once local price increases accumulate to a predefined threshold due to an irresolvable
constraint; the shadow price of that constraint would drop.
As shown below in Table 3, thirteen contingency and constraint pairs were deemed irresolvable
in 2012 and as such, had a maximum shadow price cap imposed according to the Irresolvable
Constraint Methodology. Three of the top ten real-time constraints, Odessa North 138/69 kV
transformer, China Grove to Bluff Creek 138 kV line, and Morgan Creek #1 345/138 kV
Autotransformer were designated as irresolvable in 2012. One of these constraints, the Odessa
North 138/69 kV transformer, reached the second level trigger in the procedure and had its
maximum shadow price lowered to $2,000 per MW in August 2012.
Table 3: Irresolvable Constraints
Loss of:
Base case
Graham to Long Creek
345 kV line
Odessa to Morgan Creek /
Quail 345 kV lines
Denton to Argyle / West
Denton 138 kV lines
Cabaniss to Westside
138 kV line
Key to Willow Valley
138 kV line
Key to Willow Valley
138 kV line
Odessa North to Holt
69 kV line
Overloads:
Valley Import
Bomarton to Seymour
69 kV line
Ackerly Vealmoor to Ackerly
69 kV line
Jim Crystal to West Denton
69 kV line
Naval to North Padre
69 kV line
Ackerly Vealmoor to Ackerly
69 kV line
Ackerly to Lyntegar
69 kV line
Odessa Basin to Odessa North
69 kV line
Odessa North 138/69 kV
Holt to Moss 138 kV line
transformer
Odessa Basin to Odessa North Holt to Ector Shell Tap
69 kV line
69 kV line
Odessa to Morgan Creek /
China Grove to Bluff Creek
Quail 345 kV lines
138 kV line
Sun Switch to Morgan Creek
China Grove to Bluff Creek
138 kV line
138 kV line
Morgan Creek #4
Morgan Creek #1
345 kV/138 kV
345 kV/138 kV
Autotransformer
Autotransformer
Page 42
Maximum
Shadow Price
$2,000.00
$2,000.00
$2,000.00
$2,000.00
$2,000.00
$2,554.65
$2,800.00
$2,800.00
Effective
Date
Jan 1, 2012
Jan 1, 2012
Jan 1, 2012
Jan 1, 2012
Jan 1, 2012
Jan 1, 2012
Jan 1, 2012
Jan 1, 2012
$2,274.65
$2000.00
Jan 1, 2012
Aug 6, 2012
$2,000.00
Jan 1, 2012
$2,000.00
May 3, 2012
$2,000.00
Oct 11, 2012
$2,000.00
Nov 2, 2012
ERCOT 2012 State of the Market Report
Transmission and Congestion
Figure 34 presents a slightly different set of real-time constraints. These are the most frequently
occurring. Other than the Lewisville to Jones Street constraint described previously, all
constraints in this ranking are related to wind generation. With the exception of the Odessa
North 138/69 kV transformer, which typically overloads under low wind and high local load
conditions, the other frequently occurring constraints are all related to high West zone wind. Not
only was the Odessa North 138/69 kV transformer constraint the most costly real-time constraint,
it tops the list as the most frequently occurring constraint in 2012. It was binding more than
15 percent of the time in 2012.
As the second most frequently occurring real-time constraint, West to North was binding less
than 10 percent of the time in 2012. During 2011 this constraint was the most frequently
occurring real-time constraint; active more than 20 percent of the time.
Figure 34: Most Frequent Real-Time Constraints
To maximize the economic use of scarce transmission capacity, the ideal outcome would be for
the actual transmission line flows to reach, but to not exceed the physical limits required to
maintain reliable operations. Figure 35 presents a summary of the utilization of the West to
Page 43
Transmission and Congestion
ERCOT 2012 State of the Market Report
North interface transmission constraint during 2011 and 2012. By comparing the actual flow
with the physical limit of the constraint for each dispatch interval it was binding, we can
compute its average utilization.
Figure 35: Utilization of the West to North Interface Constraint
3500
Percent Utilization
100%
90%
3000
80%
2500
70%
60%
MW
2000
Physical Limit
1500
50%
40%
Actual Flow
1000
30%
20%
500
0
10%
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
2011
0%
2012
Through July 2012, the average physical limit was approximately 2,200 MW and the average
actual flow during constrained intervals was approximately 1,900 MW. After July 2012, the
physical limit increased to an average of 2,900 MW and the actual flow increased to
approximately 2,500 MW. In March 2012, a new real-time analysis tool was implemented to
better track the dynamic nature of the transient stability limit of the West to North interface.
However, there was not a noticeable increase to the transfer limit corresponding to its
implementation due to the effects of transmission outages occurring to accommodate
maintenance activities and the installation of CREZ lines. Many of these outages were complete
by July 2012 which accounts for the increase in the West to North limit. The average annual
utilization of the West to North constraint was 87 percent in 2012, which compares favorably to
Page 44
ERCOT 2012 State of the Market Report
Transmission and Congestion
78 percent utilization experienced in 2011. Over the long term, the physical limit will continue
to increase as CREZ transmission projects are completed.
C.
Day-Ahead Constraints
In this section we review transmission constraints from the day-ahead market. To the extent the
model of the transmission system used for the day-ahead market matches the real-time
transmission system, and assuming market participants transact in the DAM similarly to how
they transact in real-time, we would expect to see the same transmission constraints appear in the
day-ahead market as actually occurred during real-time.
Figure 36: Top Ten Day-Ahead Constraints
Figure 36 presents the top ten constraints from the day-ahead market, ranked by their financial
impact as measured by congestion rent. As was the case with the real-time constraints, the
Odessa North transformer constraint had a significant financial impact, although it was not the
first on the list. However, the top three constraints are all associated with the Odessa North
substation, which further demonstrates the limiting nature of the area. The final constraint on the
list, the Moss to Amoco North Cowden 138 kV line, is similar to these Odessa North constraints.
Page 45
Transmission and Congestion
ERCOT 2012 State of the Market Report
Only the Odessa North transformer, China Grove to Bluff Creek and West to North constraints
were previously shown in the real-time list of constraints.
The next two constraints, Buttercup to Whitestone and Buda to Rutherford, are located in Central
Texas and were binding in the westward direction. They are both examples of constraints
limiting more remote generation reaching the increasing load in west Texas, especially with
certain transmission lines out of service. The Fayetteville 345/138 kV transformer constraint
limited transfers feeding 138 kV loads in Houston. The Valley import constraint had less of an
impact in 2012 than 2011 although it was active during every month.
In our final analysis of this section we review the most frequently occurring day-ahead
constraints shown in Figure 37.
Figure 37: Most Frequent Day-Ahead Constraints
Three of the constraints appearing on the list would not occur in real-time. The Eagle Pass City
to Eagle Pass DC Tie constraint appears frequently as a day-ahead constraint, but in real-time
operations all transactions with Mexico using this DC Tie are scheduled using a separate process.
The process would strictly limit the volume of transactions and not allow a constraint to occur.
Page 46
ERCOT 2012 State of the Market Report
Transmission and Congestion
The Yellowjacket to Hext and Ballinger to Humble constraints are both affected by nearby phase
shifters that depending on the tap setting of the element will have different impedances through
the phase shifter. In the day-ahead market, the phase shifters are set at one value throughout the
day, typically a mid-setting of the full range. The constraint seen in the day-ahead would likely
not bind in real-time due to the fact that the tap
settings can be changed to alter the flow over the
elements.
With the exception of the San Diego to Freer 69 kV
Figure 38: Day-Ahead Congestion Costs
ERCOT
5.8%
line located in a sparse transmission area of South
HOUSTON
3.4%
NORTH
10.5%
Texas all of the constraints listed in Figure 37 are
associated with West zone conditions.
To further emphasize the effects of West zone
congestion in 2012, Figure 38 highlights that, like
WEST
53.4%
real-time, day-ahead West zone congestion
SOUTH
27.0%
accounted for more than half the congestion in
2012.
D.
Congestion Rights Market
Congestion can be significant from an economic perspective, compelling the dispatch of highercost resources because power produced by lower-cost resources cannot be delivered due to
transmission constraint(s). Under the nodal market design, one means by which ERCOT market
participants can hedge these price differences is by acquiring Congestion Revenue Rights
(“CRRs”) between any two settlement points.
CRRs are acquired by annual and monthly auctions while Pre-assigned Congestion Revenue
Rights (“PCRRs”) are allocated to certain participants based on their historical patterns of
transmission usage. Parties receiving PCRRs pay only a fraction of the auction value of a CRR
between the same source and sink. Both CRRs and PCRRs entitle the holder to payments
corresponding to the difference in locational prices of the source and sink.
Page 47
Transmission and Congestion
ERCOT 2012 State of the Market Report
Figure 39 summarizes the revenues collected by ERCOT in each month for all CRRs, both
auctioned and allocated. These revenues are distributed to loads in one of two ways. Revenues
from cross zone CRRs are allocated to loads ERCOT wide. Revenues from CRRs that have their
source and sink in the same geographic zone are allocated to loads within that zone. This
method of revenue allocation provides a disproportionate share of CRR auction revenues to loads
located in the West zone. In 2012, CRRs with both their source and sink in the West zone
accounted for 27 percent of CRR Auction revenues. This share of revenue was allocated to West
zone loads, which accounted for only 7 percent of the ERCOT total. Allocating CRR Auction
revenues in this manner helps reduce the impact of the higher congestion on West zone prices.
Figure 39: CRR Auction Revenue
As we showed in Section I.A, Real-Time Market Prices, the annual average price for the West
zone was $38.24 per MWh, nearly $10 per MWh higher than the ERCOT-wide average. The
value of CRR Auction revenues distributed only to the West zone equated to almost $3 per MWh
more than the amounts distributed to other zones.
Page 48
ERCOT 2012 State of the Market Report
Transmission and Congestion
Next, in Figure 40 we examine the value CRR owners (in aggregate) received compared to the
price they paid to acquire the CRRs. Although results for individual participants and specific
source/sink combinations varied, we find that in most months participants did not over pay in the
auction. Across the entire year of 2012, participants spent $293 million to procure CRRs and
received $480 million.
Figure 40: CRR Auction Revenue and Payment Received
$90
2011
Auction Revenue:
Payments to CRR Owners:
$80
2012
Auction Revenue:
Payments to CRR Owners:
$370 Million
$459 Million
$293 Million
$480 Million
$70
$60
$ Millions
$50
$40
$30
$20
$10
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
CRR Payment
Feb
Jan
Dec
Nov
Oct
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
2011
Sep
Auction Revenue
$0
2012
In our next look at aggregated CRR positions, we add congestion rent to the picture. Simply put,
congestion rent is the difference between the total costs that loads pay and the total revenue that
generators receive. Congestion rent creates the source of funds used to make payments to CRR
owners. Figure 41 presents all three values for each month of 2011 and 2012. Congestion rent
for the year 2012, totaled $516 million and payments to CRR owners were $480 million. The
only months in 2012 when congestion rent was less than payments to CRR owners were January
and October through December.
Page 49
Transmission and Congestion
ERCOT 2012 State of the Market Report
Figure 41: CRR Auction Revenue, Payments and Congestion Rent
$100
2011
Auction Revenue:
Payments to CRR Owners:
Congestion Rent:
$90
$80
2012
Auction Revenue:
Payments to CRR Owners:
Congestion Rent:
$370 Million
$459 Million
$473 Million
$293 Million
$480 Million
$516 Million
$ Millions
$70
$60
$50
$40
$30
$20
$10
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Congestion Rent
Mar
Feb
Jan
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
2011
Dec
CRR Payment
Auction Revenue
$0
2012
We further analyze the relationship between congestion rent and payments to CRR owners by
separating the impacts of CRRs that are settled based on day-ahead prices from the subset of
CRRs that are paid based on real-time prices.
The top portion of Figure 42, shown below, displays the comparison of day-ahead congestion
rent to payments received by CRR owners. Congestion rent was larger than payments in most
months of 2012 and for the year congestion rent was $523 million compared to $438 million that
was paid to CRR owners.
Page 50
ERCOT 2012 State of the Market Report
Transmission and Congestion
Figure 42: Day-Ahead and Real-Time Congestion Payments and Rent
The bottom portion of Figure 42 presents a different view. For this analysis we have assumed
that all PTP Obligations have been fully funded from real-time congestion rent and any residual
real-time congestion rent is available to fund payments to the subset of CRR owners that elected
to have their CRRs be settled based on real-time prices. In 2012 there was less real-time
congestion rent than the payments to holders of PTP Obligations, resulting in a $7 million
shortfall. Further, there were real-time CRR payments of $42 Million. Hence, real-time
congestion rent was insufficient to fund all PTP Obligations and CRRs being settled in real-time
in the amount of $49 million. The next figure shows this explicitly.
Page 51
Transmission and Congestion
ERCOT 2012 State of the Market Report
Figure 43: Real-Time Congestion Payments
$120
RT CRR Payment
RT PTP Payment
RT Congestion Rent
$100
$ Millions
$80
$60
$40
2011
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
$-
Jan
$20
2012
In Figure 43 the combined payments to PTP Obligation owners and CRR owners that have
elected to receive real-time payments are compared to the total real-time congestion rent. For the
year of 2012, real-time congestion rent was $480 million, payments for PTP Obligations were
$487 million and payments for real-time CRRs were $42 million, resulting in a shortfall of
approximately $49 million for the year.
This type of shortfall typically results from discrepancies between transmission topology
assumptions used when clearing the day-ahead market and the actual transmission topology that
exists during real-time. Specifically, if the day-ahead topology assumptions allow too many
real-time congestion instruments (PTP Obligations and CRRs settled at real-time prices) with
impacts on a certain path, and that path constrains at a lower limit in real-time, there will be
insufficient rent generated to pay all of the congestion instruments.
Page 52
ERCOT 2012 State of the Market Report
Transmission and Congestion
From Figure 43 we can see that March and June through September were the months with the
most noticeable deficiencies. A detailed examination of the daily congestion pattern during
these months shows that outages of transmission facilities did occur on days with large
insufficiency.
For our last look at congestion we examine the impacts of the West to North constraint in more
detail. Figure 44 presents the price spreads between the West Hub and North load zone as
valued at three separate points in time – at the monthly CRR auction, day-ahead and in real-time.
Figure 44: West Hub to North Load Zone Price Spreads
10
8
CRR
Peak Week Day
Day Ahead
Real time
$ per MWh
6
4
2
0
(2)
(4)
(6)
10
8
Peak Week End
$ per MWh
6
4
2
0
(2)
(4)
(6)
10
Off Peak
8
$ per MWh
6
4
2
0
(2)
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
(4)
(6)
Page 53
Transmission and Congestion
ERCOT 2012 State of the Market Report
Since CRRs are sold for one of three defined time periods, weekday on peak, weekend on peak,
and off peak, Figure 44 includes a separate comparison for each.
As expected, most real-time congestion, as evidenced by the largest price spread, occurred in the
off peak period, for the months of January through June and November. The day-ahead price
spreads were very similar for this period, while the prices paid for CRRs in April were more than
the value received. Conversely, during the summer months of July and August, there was very
little congestion. In October day-ahead and real-time prices were higher at the West Hub at the
peak hours, which the results of the CRR auction did not anticipate.
Page 54
ERCOT 2012 State of the Market Report
IV.
Load and Generation
LOAD AND GENERATION
This section reviews and analyzes the load patterns during 2012 and the existing generating
capacity available to satisfy the load and operating reserve requirements. We provide specific
analysis of the large quantity of installed wind generation and conclude this section with a
discussion of the daily generation commitment process.
A.
ERCOT Loads in 2012
The changes in overall load levels from year to year can be shown by tracking the changes in
average load levels. This metric will tend to capture changes in load over a large portion of the
hours during the year. It is also important to separately evaluate the changes in the load during
the highest-demand hours of the year. Significant changes in peak demand levels play a major
role in assessing the need for new resources. They also affect the probability and frequency of
shortage conditions (i.e., conditions where firm load is served but minimum operating reserves
are not maintained). The expectation of resource adequacy is based on the value of electric
service to customers and the harm and inconvenience to customers that can result from
interruptions to that service. Hence, both of these dimensions of load during 2012 are examined
in this subsection and summarized in Figure 45.
This figure shows peak load and average load in each of the ERCOT zones from 2009 to 2012. 13
In each zone, as in most electrical systems, peak demand significantly exceeds average demand.
The North zone is the largest zone (with about 38 percent of the total ERCOT load); the South
and Houston zones are comparable (27 percent) while the West zone is the smallest (8 percent of
the total ERCOT load).
Figure 45 also shows the annual non-coincident peak load for each zone. This is the highest load
that occurred in a particular zone for one hour during the year; however, the peak can occur in
different hours for different zones. As a result, the sum of the non-coincident peaks for the zones
was greater than the annual ERCOT peak load.
13
For purposes of this analysis NOIE Load Zones have been included with the proximate geographic Load Zone.
Page 55
Load and Generation
ERCOT 2012 State of the Market Report
Figure 45: Annual Load Statistics by Zone
35
Real-Time Peak Load
Change in Real-Time Load
(2011 to 2012)
Peak
Average
ERCOT
-2.6%
-3.0%
Average Real-Time Load
30
Houston
North
South
West
Load (GW)
25
20
-4.7%
-3.8%
0.3%
2.7%
-2.2%
-4.3%
-3.1%
1.8%
15
10
5
Houston
North
South
2012
2011
2010
2009
2012
2011
2010
2009
2012
2011
2010
2009
2012
2011
2010
2009
0
West
Total ERCOT load decreased from 334 TWh in 2011 to 325 TWh in 2012, a decrease of
2.7 percent or an average of 1,130 MW every hour. Similarly, the ERCOT coincident peak
hourly demand decreased from 68,311 MW in 2011 to 66,559 MW, a decrease of 1,752 MW, or
2.6 percent. The results at the zonal level are not consistent. Average load decreased in three of
the four zones, but grew by 1.8 percent in the West zone.
To provide a more detailed analysis of load at the hourly level, Figure 46 compares load duration
curves for each year from 2009 to 2012. A load duration curve shows the number of hours
(shown on the horizontal axis) that load exceeds a particular level (shown on the vertical axis).
ERCOT has a fairly smooth load duration curve, typical of most electricity markets, with low to
moderate electricity demand in most hours, and peak demand usually occurring during the late
afternoon and early evening hours of days with exceptionally high temperatures.
Page 56
ERCOT 2012 State of the Market Report
Load and Generation
Figure 46: Load Duration Curve – All hours
70
65
2009
2010
2011
2012
60
Load (GW)
55
Frequency of Demand
> 60 GW > 50 GW > 40 GW
2,038
761
75
2,424
930
114
2,901
1,323
374
2,488
999
181
50
45
40
35
2009
30
2010
2011
25
2012
20
0
1,000
2,000
3,000
5,000
4,000
Hours
6,000
7,000
8,000
As shown in Figure 46, the load duration curve for 2012 is significantly lower than in 2011 for
approximately half of the hours in the year. This is consistent with the aforementioned
2.7 percent load decrease from 2011 to 2012.
To better illustrate the differences in the highest-demand periods between years, Figure 47 shows
the load duration curve for the five percent of hours with the highest loads. This figure also
shows that the peak load in each year is significantly greater than the load at the 95th percentile
of hourly load. From 2009 to 2012, the peak load value averaged 18 percent greater than the
load at the 95th percentile. These load characteristics imply that a substantial amount of capacity
– almost 10 GW – is needed to supply energy in less than 5 percent of the hours.
Page 57
Load and Generation
ERCOT 2012 State of the Market Report
Figure 47: Load Duration Curve – Top five percent of hours
70
68
2009
2010
2011
2012
66
Load (GW)
64
Frequency of High Demand
> 62 GW > 60 GW > 58 GW
75
160
20
114
191
51
539
374
233
181
293
96
62
60
58
56
2009
2010
54
2011
2012
52
50
0
B.
100
200
Number of Hours
300
400
Generation Capacity in ERCOT
In this section we evaluate the generation mix in ERCOT. The distribution of capacity among
the ERCOT zones is similar to the distribution of demand with the exception of the large amount
of wind capacity in the West zone. The North zone accounts for approximately 36 percent of
capacity, the South zone 29 percent, the Houston zone 21 percent, and the West zone 14 percent.
The Houston zone typically imports power, while the West zone typically exports power.
Excluding mothballed resources and including only 8.7 percent of wind capacity as capacity
available to reliably meet peak demand, the North zone accounts for approximately 39 percent of
capacity, the South zone 31 percent, the Houston zone 23 percent, and the West zone 6 percent.
Figure 48 shows the installed generating capacity by type in each of the ERCOT zones. 14
14
For purposes of this analysis, generation located in a NOIE Load Zone has been included with the proximate
geographic Load Zone
Page 58
ERCOT 2012 State of the Market Report
Load and Generation
Figure 48: Installed Capacity by Technology for each Zone
35
90
Mothballed
Private Network
Other
Hydro
Wind
Peakers - Oil or Gas
Steam - Gas
Combined Cycle - Gas
Steam - Coal
Nuclear
25
20
80
70
60
50
40
15
30
ERCOT Capacity (GW)
Installed Capacity by Zone (GW)
30
10
20
5
10
0
0
Houston
North
South
West
ERCOT
Approximately 1 GW of generation resources came online in 2012; most of which were wind
units and the remainder were solar and biomass. Even with no new natural gas units added
during 2012, natural gas generation accounts for approximately 49 percent of total ERCOT
installed capacity, and coal generation accounts for approximate 20 percent.
By comparing the current mix of installed generation capacity to that in 2007, as shown in
Figure 49, we can see the effects of longer term trends. Over these six years, wind and coal
generation are the two categories with the most increased capacity. The sizable additions in
these two categories have been more than offset by retirements of natural gas-fired steam units,
resulting in less installed capacity in 2012 than there was in 2007.
Page 59
Load and Generation
ERCOT 2012 State of the Market Report
Figure 49: Installed Capacity by Type: 2007 to 2012
100
90
80
Other
Installed Capacity (GW)
70
Hydro
Wind
60
Peakers - Oil or Gas
Mothballed
50
Steam - Gas
40
Combined Cycle - Gas
Private Network
30
Steam - Coal
Nuclear
20
10
0
2007
2012
The shifting contribution of coal and wind generation is evident in Figure 50, which shows the
percent of annual generation from each fuel type for the years 2008 through 2012. The
generation share from wind has increased every year, reaching 9 percent of the annual generation
requirement in 2012, up from 5 percent in 2008. During the same period the percentage of
generation provided by natural gas decreased from 43 percent in 2008 to 38 percent in 2010,
before increasing to 45 percent in 2012. Correspondingly, the percentage of generation produced
by coal units increased from 37 percent to 40 percent in 2010 before decreasing to 34 percent in
2012. The increase in the share of generation produced by natural gas, and corresponding
reduction in coal generation is due to historically low price of natural gas in 2012.
Page 60
ERCOT 2012 State of the Market Report
Load and Generation
Figure 50: Annual Generation Mix
100%
90%
80%
Annual Generation Mix
70%
Other
60%
Hydro
Natural Gas
50%
Wind
40%
Coal
Nuclear
30%
20%
10%
0%
2008
2009
2010
2011
2012
While coal/lignite and nuclear plants operate primarily as base load units in ERCOT, it is the
reliance on natural gas resources that drives the high correlation between real-time energy prices
and the price of natural gas fuel. There is approximately 23 GW of coal and nuclear generation
in ERCOT. Generally, when ERCOT load is above this level, natural gas resources will be on
the margin and set the real-time energy spot price. However, due to the low price of natural gas
in 2012, we observe that the share of generation produced from coal-fired and nuclear units
decreased to less than half of the energy in ERCOT, with the reduction coming from decreased
coal generation. This reduction in the share of coal generation results in an increase in the
occurrences when coal units were setting the price. This happens because the decrease in natural
gas price results in those units becoming infra-marginal; that is, less costly than the last unit
needed to satisfy total demand. As natural gas units are marginal less frequently, coal units
increasingly become marginal. We can see the results of this tradeoff in Figure 51 which shows
that the frequency with which coal was the marginal fuel was greater than 40 percent in all
months during 2012, a noticeable increase from 2011.
Page 61
Load and Generation
ERCOT 2012 State of the Market Report
Figure 51: Marginal Unit Frequency by Fuel Type
100%
90%
80%
70%
60%
50%
Wind
Gas
40%
Coal
30%
20%
10%
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
0%
2011
2012
The methodology used in this analysis reflects the details of the unit specific dispatch that are
available under the nodal market design. For every five-minute interval we determine which
units are marginal, that is they are being dispatched and their offer price is contributing to the
locational marginal price. When there is congestion, units with different prices can be marginal
at the same time. With all the marginal units identified, we aggregate by their fuel type to
compute monthly percentages. This aggregation ignores all locational price differences.
Although natural gas units continue to be marginal most of the time, the frequency at which coal
units were marginal has steadily increased since late 2011. As previously discussed this can be
attributed to the decline in natural gas prices. The frequency of wind units being marginal also
declined in 2012. This can be attributed to the reduced necessity for wind curtailments due to
transmission constraints.
Page 62
ERCOT 2012 State of the Market Report
1.
Load and Generation
Wind Generation
The amount of wind generation installed in ERCOT exceeded 10 GW by the end of 2012.
Although the large majority of wind generation is located in the West zone, there has been more
than 2 GW of wind generation recently installed in the South zone. Additionally, a private
transmission line went into service in late 2010 allowing nearly another 1 GW of West zone
wind to be delivered directly to the South zone. This section will more fully describe the
characteristics of wind generation in ERCOT.
Figure 52: Average Wind Production
6,000
5,000
Generation (MW)
4,000
3,000
2,000
1,000
Hours 1 - 4
Hours 5 - 8
Hours 9 - 12
Hours 13 - 16
Hours 17 - 20
Hours 21 - 24
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
The average profile of wind production is negatively correlated with the load profile, with the
highest wind production occurring during non-summer months, and predominately during offpeak hours. Figure 52 shows average wind production for each month in 2012, with the average
production in each month shown separately in four hour blocks. 15
15
Figure 52 shows actual wind production, which was affected by curtailments at the higher production levels.
Thus, the higher levels of actual wind production in Figure 52 are lower than the production levels that would
have materialized absent transmission constraints.
Page 63
Load and Generation
ERCOT 2012 State of the Market Report
Examining wind generation in total masks the different wind profiles that exist for locations
across ERCOT. Wind developers have more recently been attracted to the site facilities along
the Gulf coast of Texas due to the higher correlation of winds with electricity demands. Next we
compare the differences in output for wind units located in the coastal area of the South zone and
those located elsewhere in ERCOT.
Figure 53: Summer Wind Production vs. Load
3,500
Coastal
Non-Coastal
Average Wind Production (MW)
3,000
2,500
2,000
1,500
1,000
500
0
<35
35-40
40-45
45-50
50-55
ERCOT Load (GW)
55-60
60-65
>65
In Figure 53 data is presented for the summer months of June through August, comparing the
average output for wind generators located in coastal and non-coastal areas in ERCOT across
various load levels. It shows a strong negative relationship between non-coastal wind output and
increasing load levels. It further shows that the output from wind generators located in the
coastal area of the South zone is much more highly correlated with peak electricity demand.
The growing numbers of solar generation facilities in ERCOT also have an expected generation
profile highly correlated with peak summer loads. Figure 54 below compares average
summertime (June through August) hourly loads with observed output from solar and wind
Page 64
ERCOT 2012 State of the Market Report
Load and Generation
resources. Generation output is expressed as a ratio of actual output divided by installed
capacity. The solar output shown is from relatively small central station photovoltaic facilities
totaling approximately 50 MW. However, its production as a percentage of installed capacity is
the highest, exceeding 70 percent in the early afternoon, and producing more than 50 percent of
its installed capacity during peak.
Figure 54: Summer Renewable Production
100%
80%
Solar
Coastal Wind
Non Coastal Wind
Load
70%
Capacity Factor
50
40
60%
50%
30
40%
Load (GW)
90%
60
20
30%
20%
10
10%
0%
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hours of the Day
The contrast between coastal wind and non-coastal wind is also clearly displayed in Figure 54.
Coastal wind produced greater than 50 percent of its installed capacity during summer peak
hours while output from non-coastal wind was approximately 20 percent.
Page 65
Load and Generation
4,000
ERCOT 2012 State of the Market Report
Figure 55: Wind Production and Curtailment
Wind Generation
Estimated Curtailment
Wind Production & Curtailment (GWh)
3,500
3,000
2,500
2,000
1,500
1,000
500
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
0
2011
2012
Figure 55 shows the wind production and estimated curtailment quantities for each month of
2011 and 2012. This figure reveals that the total production from wind resources increased again
in 2012. More importantly, the quantity of curtailments was lower in 2012 when compared to
2011. The volume of wind actually produced was approximately 96 percent of the total available
wind in 2012, up from approximately 92 percent in 2011.
Increasing levels of wind resource in ERCOT also has important implications for the net load
duration curve faced by the non-wind fleet of resources. Net load is defined as the system load
minus wind production. Figure 56 shows the net load duration curves for selected years since
2007, normalized as a percent of peak load.
Page 66
ERCOT 2012 State of the Market Report
Load and Generation
Figure 56: Net Load Duration Curves
100%
90%
2012
Normalized Net Load
80%
2009
70%
2007
60%
50%
40%
30%
20%
0
1,000
2,000
3,000
4,000
5,000
Hours
6,000
7,000
8,000
This figure shows the continued erosion of remaining energy available for non-wind units to
serve during most hours of the year, with much less impact during the highest loads.
Even with the increased development activity in the coastal area of the South zone, more than
80 percent of the wind resources in the ERCOT region are located in west Texas. The wind
profiles in this area are such that most of the wind production occurs during off-peak hours or
other times of relatively low system demand. This profile results in only modest reductions of
the net load relative to the actual load during the hours of highest demand, but much more
significant reductions in the net load relative to the actual load in the other hours of the year.
The trend shown from 2007 in Figure 56 may continue with the addition of new wind resources
and the reduction in the curtailment of existing wind resources after completion of new
transmission facilities.
Focusing on the left side of the net load duration curve shown in Figure 57, the difference
between peak net load and the 95th percentile of net load has been between 9.5 and 12.5 GW for
the previous six years.
Page 67
Load and Generation
70
ERCOT 2012 State of the Market Report
Figure 57: Top and Bottom Ten Percent of Net Load
2012
60
2009
Net Load (GW)
50
2007
40
30
20
10
0
Hours
On the right side of the net load duration curve, the minimum net load has remained
approximately 17 GW for the past four years, even with sizable growth in total annual load. This
continues to put operational pressure on the nearly 25 GW of nuclear and coal-fired generation
currently installed in ERCOT.
Thus, although the peak net load and reserve margin requirements are projected to continue to
increase and create an increasing need for non-wind capacity to meet net load and reliability
requirements, the non-wind fleet is expected to operate for fewer hours as wind penetration
continues to increase. This outlook further reinforces the importance of efficient energy pricing
during peak demand conditions and other times of system stress, particularly within the context
of the ERCOT energy-only market design.
2.
Daily Generator Commitments
One of the important characteristics of any electricity market is the extent to which it results in
the efficient commitment of generating resources. Under-commitment can cause apparent
Page 68
ERCOT 2012 State of the Market Report
Load and Generation
shortages in real-time and inefficiently high energy prices while over-commitment can result in
excessive start-up costs, uplift charges, and inefficiently low energy prices.
This subsection evaluates the commitment patterns in ERCOT by examining the levels of excess
capacity. Excess capacity is defined as the total capacity of online plus quick-start generators
minus the demand for energy, responsive reserve, up regulation and non-spinning reserve
provided from online capacity or quick-start units. To evaluate the commitment of resources in
ERCOT, Figure 58 plots the excess capacity compared to peak load during 2012.
Figure 58: Excess On-Line and Quick Start Capacity
10
9
8
Excess Capacity (GW)
7
6
5
4
Mean = 2880 MW
3
2
1
0
30
35
40
45
50
55
60
65
70
Weekday Peak Load (GW)
The figure shows the excess capacity in only the peak hour of each weekday because the largest
generation commitment usually occurs at the peak hour. Hence, one would expect larger
quantities of excess capacity in other hours. Figure 58 shows that the excess on-line capacity
during daily peak hours on weekdays averaged 2,880 MW in 2012 which is approximately
7.8 percent of the average load in ERCOT. These values did not change significantly from 2011,
when the average excess on-line capacity was 2,901 MW, or 7.6 percent of the average load.
Even with the expected improvements in unit commitment coming from having a day-ahead
Page 69
Load and Generation
ERCOT 2012 State of the Market Report
market, if ERCOT’s day-ahead load forecast continued to show significant bias toward overforecasting peak load hours, 16 we would expect to see over commitment of generation using nonmarket means.
Figure 59: Load Forecast Error
2,500
Hours 13-16
2,000
Hours 17-20
1,500
Forecast Error (MW)
1,000
500
0
(500)
(1,000)
(1,500)
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
(2,000)
2008
2009
2010
2011
2012
From Figure 59 we can see the noticeable reduction in ERCOT’s load forecast bias since 2009.
This was due to a procedure change implemented three years ago. ERCOT now specifically
identifies and subtracts out the forecast bias and procures additional non-spin capacity in an
equal amount.
Once ERCOT assesses the unit commitments resulting from the day-ahead market, additional
capacity commitments are made, if needed, using a reliability unit commitment process that
executes both on a day-ahead and hour ahead basis. These additional unit commitments may be
16
See 2010 ERCOT SOM report at pages 49-51 and 2009 ERCOT SOM report at pages 68-70.
Page 70
ERCOT 2012 State of the Market Report
Load and Generation
made for one of two reasons. Either additional capacity is required to ensure forecasted total
demand will be met, or a specific generator is required to resolve transmission congestion.
Figure 60 summarizes, by month, the number of hours with units committed via the reliability
unit commitment process. We observe a significant reduction in the reliance upon the reliability
unit commitment process in 2012 as compared to 2011. Approximately one third of the hours
during 2011 had at least one unit committed by ERCOT through the reliability unit commitment
process. The amount of time during 2012 reduced to three percent. The primary reason for the
reduction is likely the less extreme weather and resulting lower load levels experienced during
2012. Lower loads resulted in reduced congestion and therefore reduced need for specific units
to be brought online to resolve.
Figure 60: Frequency of Reliability Unit Commitments
Hours of Reliability Unit Committment
600
500
400
300
200
100
2011
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
Dec
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
0
2012
There was an operational change midway through 2011 which also contributed to the reduced
frequency of reliability unit commitments. During the initial months of operating the nodal
Page 71
Load and Generation
ERCOT 2012 State of the Market Report
market it was common for ERCOT to commit units that were providing non-spin reserves if they
were needed to resolve congestion. This practice was greatly reduced starting in July 2011.
The next analysis compares the average dispatched output of the reliability committed units with
their operational limits. Figure 61 shows that for most months of 2012 the amount of capacity
actually dispatched from units brought on line via the reliability unit commitment process was
less than the 200 to 300 MW that was typical for 2011.
Figure 61: Reliability Unit Commitment Capacity
1000
Average High Capacity Limit
Average Dispatch Level
900
800
700
MW
600
500
400
300
200
100
2011
Dec
Nov
Oct
Sep
Aug
Jul
Jun
Apr
May
Mar
Feb
Jan
Dec
Nov
Oct
Sep
Aug
Jul
Jun
Apr
May
Mar
Feb
Average Low Capacity Limit
Jan
0
2012
The notable exception was in April 2012, when the large amounts of reliability unit committed
capacity were related to brief generator outages resulting in reactive power deficiencies in the
Dallas-Fort Worth area. This was similar to the situation that existed during October 2011. The
larger quantity of committed capacity in February 2011 was a result of ERCOT operator action
taken to attempt to ensure overall capacity adequacy during both the extreme cold weather event
that occurred early in the month, and a subsequent bout of cold weather that occurred one week
later.
Page 72
ERCOT 2012 State of the Market Report
V.
Resource Adequacy
RESOURCE ADEQUACY
One of the primary functions of the wholesale electricity market is to provide economic signals
that will facilitate the investment needed to maintain a set of resources that are adequate to
satisfy the system’s demands and reliability needs. We begin this section with an evaluation of
these economic signals by estimating the “net revenue” new resources would receive from the
markets. Next, our review of the effectiveness of the Public Utility Commission’s Scarcity
Pricing Mechanism includes two recommendations for market design improvements. We
conclude this section with a review of the contributions from demand response toward meeting
resource adequacy objectives in ERCOT and our third recommended improvement.
A.
Net Revenue Analysis
Net revenue is the total revenue that can be earned by a new generating unit less its variable
production costs. Put another way, it is the revenue in excess of short-run operating costs that is
available to recover a unit’s fixed and capital costs, including a return on the investment. Net
revenues from the energy and ancillary services markets together provide the economic signals
that inform suppliers’ decisions to invest in new generation or retire existing generation. In longrun equilibrium, markets should provide sufficient net revenue to allow an investor to receive a
return of, and on an investment in a new generating unit that is needed. In the short-run, if the
net short-run revenues produced by the market are not sufficient to justify entry, then one or
more of these conditions exist:
•
New capacity is not needed because there is sufficient generation already available;
•
Load levels, and thus energy prices, are temporarily low due to mild weather or economic
conditions;
•
Market rules or operational practices are causing revenues to be reduced inefficiently; or
•
Market rules are not sufficiently linked in short-term operations to ensure long-term
resource adequacy objectives are met.
Likewise, the opposite would be true if the markets provide excessive net revenues in the shortrun. The persistence of excessive net revenues in the presence of a capacity surplus is an
Page 73
Resource Adequacy
ERCOT 2012 State of the Market Report
indication of competitive issues or market design flaws. In this section, we analyze the net
revenues that would have been received by various types of generators in each zone.
Although most suppliers are likely to receive the bulk of their revenues through bilateral
contracts, the spot prices produced in the real-time energy market should drive the bilateral
energy prices over time and are appropriate to use for this evaluation. For purposes of this
analysis, heat rates of 7 MMBtu per MWh for a combined cycle unit, 10.5 MMBtu per MWh for
a combustion turbine, and 9.5 MMBtu per MWh for a new coal unit were assumed. Variable
operating and maintenance costs of $4 per MWh for the natural gas units and $5 per MWh for
the coal unit and fuel and variable operating and maintenance costs of $8 per MWh for the
nuclear unit were assumed. For purposes of this analysis, a total outage rate (planned and
forced) of 10 percent was assumed for each technology.
The energy net revenues are computed based on the generation weighted settlement point prices
from the real-time energy market. Weighting the energy values in this way masks what may be
very high locational values for a specific generator location. Some generators may also receive
uplift payments because of their specific reliability contributions, either as a reliability must run,
or through the reliability unit commitment. This source of revenue is not considered in this
analysis. The analysis also includes simplifying assumptions that can lead to over-estimates of
the profitability of operating in the wholesale market. The following factors are not explicitly
accounted for in the net revenue analysis: (i) start-up costs, which can be significant; and
(ii) minimum running times and ramp restriction, which can prevent the natural gas generators
from profiting during brief price spikes. Despite these limitations, the net revenue analysis
provides a useful summary of signals for investment in the wholesale market.
Figure 62 shows the results of the net revenue analysis for four types of hypothetical new units in
2011 and 2012. These are: (a) natural gas-fired combined-cycle, (b) natural gas-fired
combustion turbine, (c) coal-fired generator, and (d) a nuclear unit. For the natural gas-fired
technologies, net revenue is calculated by assuming the unit will produce energy in any hour for
which it is profitable and by assuming it will be available to sell reserves and regulation in other
hours that it is available (i.e., when it is not experiencing a planned or forced outage). For coal
Page 74
ERCOT 2012 State of the Market Report
Resource Adequacy
and nuclear technologies, net revenue is calculated by assuming that the unit will produce at full
output.
Figure 62: Estimated Net Revenue by Zone and Unit Type
$300
Regulation
Reserves
Energy Sales
Net Revenue ($ per kW-yr)
$250
$200
$150
$100
$50
CC
CT
Coal
Nuke
2011
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
Houston
North
South
West
$-
CC
CT
Coal
Nuke
2012
Figure 62 shows that the net revenue for every generation technology type decreased
substantially in 2012 compared to each zone in 2011.
•
For a new coal unit, the estimated net revenue requirement is approximately $210 to
$270 per kW-year. The estimated net revenue in 2012 for a new coal unit was
approximately $35 per kW-year.
•
For a new nuclear unit, the estimated net revenue requirement is approximately $280 to
$390 per kW-year. The estimated net revenue in 2012 for a new nuclear unit was
approximately $134 per kW-year.
Prior to 2005, net revenues were well below the levels necessary to justify new investment in
coal and nuclear generation. Higher natural gas prices through 2008 allowed energy prices to
remain at levels high enough to support new entry for these technologies. The production costs
Page 75
Resource Adequacy
ERCOT 2012 State of the Market Report
of coal and nuclear units did not change significantly over this period, leading to a dramatic rise
in net revenues. However, since 2008 natural gas prices have been on the decline, resulting in
reduced net revenues for coal and nuclear technologies. Even with the higher energy prices
experienced in 2011, net revenues for these technologies were insufficient to support new entry.
With the further decline in natural gas prices and few occurrences of shortage pricing, the
estimated net revenue for either a new coal or a nuclear unit in ERCOT was well below the
levels required to support new entry in 2012.
The next two figures provide an historical perspective of the net revenues available to support
new gas turbine (Figure 63) and combined cycle generation (Figure 64).
Figure 63: Gas Turbine Net Revenues
$140
Reserves
Energy Sales
Net Revenues ( $ per kW - year)
$120
$100
$80
$60
$40
$20
2007
2008
2009
2010
2011
South
North
Houston
South
North
Houston
South
North
Houston
South
North
Houston
South
North
Houston
South
North
Houston
$0
2012
Based on our estimates of investment costs for new units, the net revenue required to satisfy the
annual fixed costs (including capital carrying costs) of a new gas turbine unit ranges from $80 to
Page 76
ERCOT 2012 State of the Market Report
Resource Adequacy
$105 per kW-year. The estimated net revenue in 2012 for a new gas turbine was approximately
$25 per kW-year, far below the levels required to support new gas turbine generation.
For a new combined cycle unit, the estimated net revenue requirement is approximately $105 to
$135 per kW-year. The estimated net revenue in 2012 for a new combined cycle unit was
approximately $42 per kW-year, also far below the levels to support new combined cycle
generation in ERCOT.
Figure 64: Combined Cycle Net Revenues
$250
Regulation
Reserves
Energy Sales
Net Revenues ($ per kW - Year)
$200
$150
$100
$50
2007
2008
2009
2010
2011
South
North
Houston
South
North
Houston
South
North
Houston
South
North
Houston
South
North
Houston
South
North
Houston
$-
2012
Even though net revenues for the Houston and South zones in 2008 may have appeared to be
sufficient to support new natural gas-fired generation, it was actually extremely inefficient
transmission congestion management and inefficient pricing mechanisms associated with the
deployment of non-spinning reserves which led to high prices and resulting higher than
warranted net revenues. Discounting the effect that the 2008 results would have had on forward
Page 77
Resource Adequacy
ERCOT 2012 State of the Market Report
price signals, we find that 2011 was the first time in five years that net revenues have been
sufficient to support either new gas turbine or combined cycle generation.
These results indicate that the ERCOT markets would not have provided sufficient revenues to
support profitable investment in any of the types of generation technology evaluated. Higher
energy prices in the West zone during 2012 resulted in higher net revenues in that zone, but they
were still not high enough to support new entry there. The net revenues in 2012 were much
lower than in 2011. However, it is important to recognize that 2011 was highly anomalous, with
some of the hottest summer temperatures on record. Net revenues may have been sufficient to
cover the costs of a new combined cycle or new combustion turbine in 2011, however, we would
not expect this to be consistently true in years with comparable reserve margins absent the
extreme weather conditions, as evidenced by the 2012 net revenue results.
To provide additional context for the net revenue results presented in this section, we also
compared the net revenue in the ERCOT market for two types of natural gas-fired technologies
with the net revenue that those technologies could expect in other wholesale markets with
centrally cleared capacity markets. The technologies are differentiated by their assumed heat
rate; 7,000 MMBtu per MWh for combined cycle and 10,500 MMBtu per MWh for simple-cycle
combustion turbine.
Figure 65 compares estimates of net revenue for the ERCOT North zone, PJM, two locations
within the New York ISO, and the Midwest ISO. The figure includes estimates of net revenue
from energy, reserves and regulation, and capacity. ERCOT does not have a capacity market,
and thus, does not have any net revenue from capacity sales. Figure 65 shows that net revenues
increased from 2011 to 2012 for both technologies in NY ISO. In PJM net revenue decreased
and in Midwest ISO net revenues remained flat. In the figure below, net revenues are calculated
for central locations. However, there are load pockets within each market where net revenue and
the cost of new investment may be higher. Thus, even if new investment is not generally
profitable in a market, it may be economic in certain areas. Finally, resource investments are
driven primarily by forward price expectations, so historical net revenue analyses do not provide
a complete picture of the future pricing expectations that will spur new investment.
Page 78
ERCOT 2012 State of the Market Report
Resource Adequacy
Figure 65: Comparison of Net Revenue of Gas-Fired Generation between Markets
$250
$ per kW-year
$200
Capacity
Ancillary Services
Energy
$150
$100
$50
B.
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
7,000
10,500
$-
'09 '10 '11 '12
'09 '10 '11 '12
'09 '10 '11 '12
'09 '10 '11 '12
'10 '11 '12
ERCOT
PJM
NYISO (Capital)
NYISO (NYC)
MISO
Effectiveness of the Scarcity Pricing Mechanism
The Public Utility Commission of Texas (“PUCT”) adopted rules in 2006 that define the
parameters of an energy-only market. These rules include a Scarcity Pricing Mechanism
(“SPM”) that relaxed the existing system-wide offer cap by increasing it in multiple steps until it
reached $3,000 per MWh shortly after the implementation of the nodal market. PUCT SUBST. R.
25.505 provides that the IMM may conduct an annual review of the effectiveness of the SPM.
This subsection provides an assessment of the effectiveness of the SPM in 2012 under ERCOT’s
energy-only market structure.
Experiencing reduced levels of generation development activity coupled with higher than
expected loads resulting in diminishing planning reserve margins, the PUCT has devoted
considerable effort over the past two years deliberating issues related to resource adequacy.
These deliberations have included the question of whether that planning reserve margin is a
Page 79
Resource Adequacy
ERCOT 2012 State of the Market Report
target or a minimum requirement. Further, if it is a minimum requirement, whether the energyonly market design can ensure the desired reliability level or whether an alternate market design
mechanism may be required. To date, the PUCT has taken no action to change the fundamental
energy-only nature of the ERCOT market or the designation of the planning reserve margin as a
target. However, there have been changes to the rules governing the system-wide offer cap and
peaker net margin mechanism.
Approved during 2012, new PUCT SUBST. R. 25.508 increased the system-wide offer cap to
$4,500 per MWh effective August 1, 2012. As shown in Figure 15 on page 15, there was only a
brief period when energy prices rose to the cap after this change was implemented. Revisions
were also adopted to PUCT SUBST. R. 25.505 which specified future increases to the systemwide offer cap as follows:
•
•
•
$5,000 per MWh beginning on June 1, 2013,
$7,000 per MWh beginning on June 1, 2014, and
$9,000 per MWh beginning on June 1, 2015.
The SPM includes a provision termed the Peaker Net Margin (“PNM”) that is designed to
measure the annual net revenue of a hypothetical peaking unit. This aspect of the rule was also
amended in 2012. Under the current rule, if the PNM for a year reaches a cumulative total of
$300,000 per MW, 17 the system-wide offer cap is then reduced to the higher of $2,000 per MWh
or 50 times the daily natural gas price index. 18 Figure 66 shows the cumulative PNM results for
each year from 2006 through 2012 and shows that PNM in 2012 was the lowest it has been since
its implementation.
17
18
The proxy combustion turbine in the Peaker Net Margin calculation assumes a heat rate of 10 MMBtu per MWh
and includes no other variable operating costs or startup costs.
These values were increased from a previous threshold of $175,000 per MW and an LCAP of $500 per MWh.
Page 80
ERCOT 2012 State of the Market Report
Resource Adequacy
Figure 66: Peaker Net Margin
As previously described, the net revenue required to satisfy the annual fixed costs (including
capital carrying costs) of a new gas turbine unit ranges from $80,000 to $105,000 per MW-year.
Thus, as shown in Figure 66 and consistent with the previous findings in this section relating to
net revenue, the PNM was nowhere near sufficient to support new entry in 2012. Only in two of
the seven years since the rule was implemented has the PNM been sufficient – 2008 and 2011.
A significant portion of the net revenue increase in 2008 was associated with extremely
inefficient transmission congestion management and inefficient pricing mechanisms associated
with the deployment of non-spinning reserves. 19 With these issues addressed in the zonal
market, the PNM dropped substantially in 2009 and 2010. The extreme weather experienced in
2011 was highly anomalous. Hence, although the PNM may have been sufficient to cover the
costs of a new combustion turbine in 2011, we would not expect this to be true on a continuous
basis into the future.
19
See 2008 ERCOT SOM Report at pages 81-87.
Page 81
Resource Adequacy
ERCOT 2012 State of the Market Report
Shortage Pricing and Resource Adequacy
Efficient electricity markets allow energy prices to rise substantially at times when the available
supply is insufficient to simultaneously meet both energy and minimum operating reserve
requirements. Ideally, energy and reserve prices during shortages should reflect the diminished
system reliability under these conditions, which is equal to the increased probability of “losing”
load times the value of the lost load. Allowing energy prices to rise during shortages mirrors the
outcome expected if loads were able to actively specify the quantity of electricity they wanted
and the price they would be willing to pay. The energy-only market design relies exclusively on
these relatively infrequent occurrences of high prices to provide the appropriate price signal for
demand response and new investment when required. In this way, energy-only markets can
provide price signals that will sustain a portfolio of resources to be used in real time to satisfy the
needs of the system. However, this portfolio may produce a planning reserve margin that is less
than the planning reserve target, which is discussed in the next subsection.
The expectation of competitive energy market outcomes is no different in energy-only than in
markets that include a capacity market. However, capacity markets are designed to ensure a
specified planning reserve margin, which may be higher than what an energy-only market would
achieve. Under this condition the higher planning reserve margin will serve to reduce the
frequency of shortages in the energy market.
As a general principle, competitive and efficient market prices should be consistent with the
marginal cost of the marginal action taken to satisfy the market’s demand. In the vast majority
of hours, the marginal cost of the marginal action is associated with the dispatch of the last
generator required to meet demand. It is appropriate and efficient in these hours for this
generator to “set the price”. However, this is not true under shortage conditions. When the
system is in shortage, the demand for energy and minimum operating reserves cannot be satisfied
with the available resources, which will cause the system operator to take one or more of the
following actions:
•
Sacrifice a portion of the operating reserves by dispatching them for energy;
•
Voluntarily curtail load through demand response programs;
•
Curtail exports or make emergency imports; or
•
Involuntarily curtail load.
Page 82
ERCOT 2012 State of the Market Report
Resource Adequacy
A market design that adheres to the pricing principles stated above will set prices that reflect
each of these actions. When there is a shortage of supply in the market, the marginal action first
taken by the system operator is generally to not satisfy operating reserves requirements (i.e.,
dispatching reserves for energy). Diminished operating reserves results in diminished reliability,
which has a real cost to electricity consumers. In this case, the value of the foregone reserves –
which is much higher than the marginal cost of the most expensive online generator – should be
reflected in energy prices to achieve efficient economic signals governing investment in
generation, demand response and transmission.
With the implementation of the nodal market, more reliable and efficient shortage pricing has
been achieved by establishing pricing rules that recognize when operating reserve shortages exist
and allowing energy prices to rise automatically. Figure 16 on page 16 clearly shows this
relationship between increasing prices as operating reserve levels decline. This approach is more
reliable than what existed in the previous zonal market because it is not dependent upon the
submission of high-priced offers by small market participants to be effective. It is also more
efficient during the vast majority of time in which shortage conditions do not exist because it is
not necessary for market participants to effectively withhold lower cost resources by offering
relatively small quantities at prices dramatically higher than their marginal cost. At times when
there is insufficient capacity available to meet both energy and minimum operating reserve
requirements, all available capacity will be dispatched and the clearing price will rise in a
predetermined manner to a maximum of the system-wide offer cap.
Although the nodal market implementation brought about more reliable and efficient shortage
pricing, there remain aspects of the ERCOT real-time energy pricing that can be improved. As
discussed later in this section, prices during the deployment of load resources do not reflect the
value of reduced reliability which occurs when responsive reserves have been converted to
energy.
Similarly, during the first year of nodal market operation when non-spinning reserves were
deployed (converted to energy), prices rarely reflected the marginal cost of the action being
taken. Non-spinning reserves are provided primarily by off-line natural gas-fired combustion
turbines capable of starting in 30 minutes or less. The implementation of the nodal market
Page 83
Resource Adequacy
ERCOT 2012 State of the Market Report
significantly increased market efficiencies in a number of areas, including the move to a five
minute rather than 15 minute energy dispatch. However, it lacked an efficient economic
commitment mechanism for resources such as offline gas turbines and other resources that are
not immediately dispatchable in the five minute energy dispatch. This led to prices that were
inefficiently low because they did not represent the costs associated with starting and running the
gas turbines that were being deployed to meet demand.
The changes described in NPRRs 426 and 428 were implemented in early 2012, and added
requirements for providers of non-spinning reserve to make that capacity available to ERCOT’s
dispatch software, subject to certain offer floors. 20 Providers are now able to specify the price at
which they are willing to convert their non-spinning reserve capacity to energy. Further,
ERCOT uses this price information to determine which non-spin units to deploy. Real-time
energy price formation has been improved, but the current mechanism is sub-optimal from a
reliability and efficiency perspective. We continue to recommend that ERCOT develop a
mechanism that will rationally commit generation that can start or load resources that can curtail
within 30 minutes.
This deficiency in ERCOT’s nodal market design should be addressed by implementing “look
ahead” dispatch functionality for the real-time energy market to produce energy and ancillary
services commitment and dispatch results that are co-optimized and recognize anticipated
changes in system demands. 21 This additional functionality represents a major change to
ERCOT systems; one we recommend together with improved pricing provisions that will allow
offers from load resources to set prices if they are required to meet system demand.
Effective look ahead dispatch functionality should also reduce the price dampening effects of
energy produced by units operating below their low sustainable operating limit. Although
alternatives have been suggested to address this issue in a standalone manner, we believe the
better approach will be to develop a comprehensive look ahead dispatch solution.
20
The offer floors for online and offline non-spinning reserves are $120 and $180 per MWh, respectively.
NPRR427 requires that energy offers from generation resources providing responsive reserve and up regulation
reserves be priced at the system-wide offer cap.
21
See Direct Testimony of David B. Patton, PUCT Docket No. 31540, (Nov. 10, 2005), at pages 35-41.
Page 84
ERCOT 2012 State of the Market Report
Resource Adequacy
Aside from the offer floors for non-spinning reserves, the Power Balance Penalty Curve
(“PBPC”) and the offer floors for up regulation and responsive reserve provided from generation
resources defines the relationship between the quantity of operating reserve deficiency and the
resulting energy price. The PBPC was modified during 2012 in conjunction with the increase in
the system-wide offer cap to $4,500 per MWh. Figure 67 compares the original PBPC in place
at the start of the nodal market and the current curve, as modified in 2012.
Figure 67: Power Balance Penalty Curves
$5,000
$4,500
$4,000
$ per MWh
$3,500
$3,000
$2,500
$2,000
$1,500
Original
$1,000
Current
$500
$0
0
100
200
MW
300
400
Under the current curve, if operating reserves are deficient by 5 MW or less, the energy price
will be $250 per MWh. If the deficiency is greater than 150 MW but less than 200 MW, the
energy price would be set at $4,000 per MWh. Once the 200 MW from the PBPC is exhausted,
the only remaining energy available is from generator provided responsive reserves and up
regulation reserves. Since energy provided by these services is required to be offered at the
system-wide offer cap, real-time energy prices will be set at that level.
Page 85
Resource Adequacy
ERCOT 2012 State of the Market Report
Table 4: Power Balance Penalty Curve
Maximum
Operating Reserve
Deficiency
(MW)
1
5
10
20
30
40
50
>50
100
150
200
>200
Energy Price
($ per MWh)
Energy Price
($ per MWh)
Original Curve
$200
$250
$300
$400
$500
$1,000
$2,250
$3,001
Current Curve
$250
$300
$400
$500
$1,000
$2,250
$3,000
$3,500
$4,000
$4,501
The current relationship between operating reserve deficiency and energy prices defined by the
PBPC and the operating reserve offer floors has no real analytic basis other than having its end
anchored by the system-wide offer cap. The intermediate values are set at values acceptable to
the collective agreement of stakeholders. A more analytically rigorous approach would be to
introduce an operating reserve pricing mechanism that reflects the operational loss of load
probability (“LOLP”) at varying levels of operating reserves multiplied by the value of lost load
(“VOLL”). The LOLP would be equal to 1.0 when operating reserves fall to the level where
involuntary load shedding is directed by ERCOT. The LOLP would decline exponentially from
1.0 as the level of operating reserves increased.
The implementation of such a curve is currently being evaluated under different approaches.
The most complex approach would be to implement real-time co-optimization of energy and
reserves with an operating reserve demand curve. The approach named “Solution B+” is likely
to be easier to implement and would introduce the operating reserve demand curve but not
include real-time co-optimization.22 And yet another approach would be to adjust the current
22
See ERCOT Presentation Regarding Potential Implementation of Scarcity Pricing Proposal Offered by Professor
Hogan, PUCT Docket No. 40000, (Jan. 22, 2013).
Page 86
ERCOT 2012 State of the Market Report
Resource Adequacy
operating reserve offer floors to better reflect LOLP * VOLL at various levels of operating
reserves. Each of these approaches could result in more efficient pricing of operating reserve
shortages. However, for any of these approaches to result in planning reserve margins over the
long-term that meet the historical standard of one loss of load event in ten years would likely
require an increase to the level of minimum operating reserves significantly beyond what is
required to maintain reliable system operations, or by establishing shortage pricing that is
substantially higher than the system-wide offer cap rising to $9,000 per MWh or most reasonable
estimates of VOLL. As discussed below, each of these changes can introduce costly operational
inefficiencies into the ERCOT energy markets.
As the system-wide offer cap increases to $5,000 per MWh and beyond, it is likely under the
current market mechanisms that prices will rise to levels approaching the VOLL when the
available reserves are at levels where the LOLP is much less than 1.0 and involuntary load
curtailment is not imminent. We have two recommendations to address this concern. The first is
to modify the slope of the existing PBPC and the offer floor for responsive reserve service to
provide a more gradual slope up to the system-wide offer cap. This could be accomplished by
any of three approaches discussed above. We also recommend modifying the automatic pricing
of unoffered capacity such that it is not all priced at the system-wide offer cap to avoid the
inefficiencies associated with the automated economic withholding of such capacity.
Shortage Pricing, Capacity Markets, and Resource Adequacy
Regardless of the means by which revenues are produced in a wholesale electricity market, it is
fundamental that investment will only occur when the total net revenues expected by the investor
are greater than its entry costs (including profit on its investment). Additionally, these sources of
revenue must be available to all resources, both new and existing, in order to facilitate efficient
investment, maintenance, and retirement decisions by all suppliers.
In an energy only market, the primary source of such revenue is the net revenues received during
periods of shortage. Expectations about both the magnitude of the energy price during shortage
conditions and the frequency of shortage conditions are the primary means to attract new
investment in an energy-only market. If the expected revenues are not high enough to facilitate
enough investment to satisfy the planning reserve target, one option is to increase the shortage
Page 87
Resource Adequacy
ERCOT 2012 State of the Market Report
pricing levels to levels that substantially exceed the expected value of lost load. As the planning
reserve levels grow, however, the frequency of shortages will tend to drop sharply, which can
make it difficult to use this means to meet planning reserve targets. 23 Additionally, as discussed
below, such approaches introduce costly operational inefficiencies in to the ERCOT energy
markets.
Most other competitive electricity markets do not rely solely on shortage pricing to generate
sufficient revenue to support the capacity additions necessary to satisfy their planning reserve
requirements. They employ capacity markets to competitively generate capacity payments over
the year that are made to suppliers in return for meeting defined capacity obligations. Capacity
prices and associated payments vary monthly or annually based on long-term planning reserve
levels, independent of the real-time supply and demand conditions. These capacity markets are
designed to ensure that a specified planning reserve margin is achieved.
In 2012, ERCOT engaged The Brattle Group to assess its resource adequacy outlook by
evaluating a number of market design scenarios. 24 Brattle also supplemented its report with the
following table that presents a comparison of costs and reliability for the energy-only market and
two capacity market scenarios with 10% and 14% reserve margin requirements. 25 The results of
this analysis are summarized in the table below.
Brattle estimates that even with $9,000 per MWh system-wide offer caps, economic equilibrium
for the ERCOT energy-only market is achieved at an 8 percent planning reserve margin,
although the actual reserve margin outcomes will be uncertain. Brattle further estimates that in
the energy-only market at annual equilibrium, wholesale generation costs will be $18.3 billion.
In contrast, Brattle’s assessment of a capacity market with a more certain 14 percent reserve
margin expectation, estimates generation costs at annual equilibrium to be $18.7 billion.
23
The difficulty of relying primarily on shortage pricing will depend on how high the planning reserve target
is relative to the planning reserve levels any energy-only market priced at the expected value of lost load
would provide. See the discussion of the Brattle Report below.
24
ERCOT Investment Incentives and Resource Adequacy, The Brattle Group, PUCT Docket No. 37987 (June 1,
2012).
25
Customer Cost Comparison, The Brattle Group, PUCT Docket No. 40000 (Sept. 4, 2012).
Page 88
ERCOT 2012 State of the Market Report
Resource Adequacy
It is important to recognize that this increase in cost is not due to the introduction of the capacity
market, it is due to the requirement to sustain a planning reserve margin greater than 8 percent.
In fact, the Brattle analysis indicates that a capacity market would deliver the higher planning
reserve margin at a relatively low incremental cost with much more certainty.
Recent studies have indicated that to maintain the same small level of risk of having an
involuntary curtailment of firm load, the planning reserve target should be increased from
13.75 percent to approximately 16 percent. Hence, the difficulty of satisfying ERCOT’s
planning needs with shortage pricing alone will grow if this recommendation is adopted. Shown
below in Figure 68 is ERCOT’s most current projection of reserve margins. It indicates that the
region will have a 13.2 percent reserve margin heading into the summer of 2013. With the
addition of recently announced generation additions, in 2014 the reserve margin is expected to
reach 13.8 percent -- just barely above the current target. The bulk of the new capacity being
added is natural gas-fired generation, approximately a quarter of which is expansions at existing
facilities.
Page 89
Resource Adequacy
ERCOT 2012 State of the Market Report
Figure 68: Projected Reserve Margins
16%
Existing Capacity
New Coal
New Gas
New Other
New Wind (8.7% of installed capacity)
Target Reserve Margin: 13.75%
14%
Projected Reserve Margin
12%
10%
8%
6%
4%
2%
0%
2013
2014
2015
2016
2017
2018
Source: ERCOT Capacity Demand Reserve Reports / 2013 data from Winter 2012, 2014 - 2018 from May 2013
In response to these observations, proposals have been put forth that would introduce significant
operational inefficiencies into the ERCOT energy markets, such as a requirement to substantially
increase the quantity of operating reserves ERCOT procures and to, by rule, economically
withhold these surplus reserves from the market. Such approaches would introduce significant
inefficiencies into ERCOT day ahead and real time operations in an effort to manufacture more
frequent shortage pricing and a higher planning reserve margin than would be achieved in a pure
energy-only market framework. However, such approaches will not guarantee that the planning
reserve targets will be satisfied and, because of the resulting inefficiencies, will be more costly
for ERCOT’s consumers. Hence, consistent with Brattle’s findings, it is our view that, if the
planning reserve margin is viewed as a minimum requirement, implementation of a capacity
market is the most efficient mechanism to achieve this objective.
As observed by Brattle, a
well-designed capacity market can efficiently meet a planning reserve requirement without
Page 90
ERCOT 2012 State of the Market Report
Resource Adequacy
impairing the efficiency of energy market operations. However, there are many determinations
required in the design, implementation and maintenance of a capacity market construct. 26
C.
Demand Response Capability
Demand response is a term that broadly refers to actions that can be taken by end users of
electricity to reduce load in response to instructions from ERCOT or in response to certain
market or system conditions. The ERCOT market allows participants with demand-response
capability to provide energy and reserves in a manner similar to a generating resource. The
ERCOT Protocols allow for loads to actively participate in the ERCOT administered markets as
Load Resources. Additionally, loads may participate passively in the market by simply adjusting
consumption in response to observed prices. Unlike active participation in ERCOT administered
markets, passive demand response is not directly tracked by ERCOT.
ERCOT allows qualified load resources to offer responsive reserves and non-spinning reserves
into the day-ahead ancillary services markets. Those providing responsive reserves must have
high set under-frequency relay equipment, which enables the load to be automatically tripped
when the frequency falls below 59.7 Hz, which will typically occur only a few times in each
year. Deployments of non-spinning reserves occur much more frequently. To date, load
resources have shown a clear preference for providing responsive reserve service.
As of December 2012, approximately 2,500 MW of capability were qualified as Load Resources.
Figure 69 shows the amount of responsive reserves provided from load resources on a daily basis
in 2012. The high level of participation by demand response in the ancillary service markets sets
ERCOT apart from other operating electricity markets. For reliability reasons the maximum
amount of responsive reserves that can be provided by load resources was limited to 1,150 MW
until April 2012. At that time, the limitation on load resources providing responsive reserve
increased to 1,400 MW, corresponding with the increase in total responsive reserve
requirements.
26
ERCOT Investment Incentives and Resource Adequacy, The Brattle Group, at 115-119, PUCT Docket No. 37987
(June 1, 2012).
Page 91
Resource Adequacy
ERCOT 2012 State of the Market Report
Figure 69: Daily Average of Responsive Reserves provided by Load Resources
Figure 69 shows that it took a few months after implementing the increased requirement for the
amount of offers by load resources to routinely reach this level. Notable exceptions include a
prolonged reduction from November 2008 through January 2009 that was likely a product of the
economic downturn and its effect on industrial operations. Seasonal reductions were also
observed during late 2009 and 2012.
During 2011 there was a significant reduction in loads offering to provide responsive reserve
during early February and again starting in mid-July. Both of these times corresponded with
expected high real-time prices. Since load resources provide capacity by reducing their
consumption, they have to actually be consuming energy to be eligible to provide the capacity
service. During periods of expected high prices the price paid for the energy can exceed the
value received from providing responsive reserves.
Page 92
ERCOT 2012 State of the Market Report
Resource Adequacy
Pricing During Load Deployments
During times when there are shortages of supply offers available for dispatch and Responsive
Reserves are deployed, that is, converted to energy as one of the last steps taken before shedding
firm load, the value of the foregone reserves – which is much higher than the marginal cost of
the most expensive online generator – should be reflected in energy prices to achieve efficient
economic signals governing investment in generation, demand response and transmission.
Unfortunately, ERCOT’s dispatch software does not recognize that load has been curtailed, and
computes prices based on supplying only the remaining load. A good example of this situation
occurred on August 4, 2011. Figure 70 displays available reserves and the system price for that
afternoon and shows that even though reserves were below required levels, system price dropped
to $60 per MWh. At this level prices are being set based on supply offers and do not reflect the
value of the load that is being curtailed to reliably serve the remaining system demand.
Figure 70: Pricing During Load Deployments
3500
Reserves & Load (MW) ---- $ per MWh for System Price
Actual Reserves
System Price
3000
2500
2000
1500
1000
Required Reserve
System
Price drops
when
Reserves
are below
required
levels due
to Load
deployment
500
Total Load Deployed
0
8/4 12:00 8/4 13:00 8/4 14:00 8/4 15:00 8/4 16:00 8/4 17:00 8/4 18:00 8/4 19:00
We recommend that ERCOT implement system changes that will ensure that all demand
response that is actively deployed by ERCOT be incorporated into the dispatch software so that
Page 93
Resource Adequacy
ERCOT 2012 State of the Market Report
such deployments will be able to set the price at the value of load at times when such
deployments are necessary to reliably serve the remaining system demand. This includes load
resources and Emergency Response Service (ERS) providers being deployed for the services
they contracted to provide or when firm load is involuntarily curtailed.
Page 94
ERCOT 2012 State of the Market Report
VI.
Analysis of Competitive Performance
ANALYSIS OF COMPETITIVE PERFORMANCE
In this section we evaluate market power from two perspectives, structural (does market power
exist) and behavioral (have attempts been made to exercise it). We examine market structure by
using a pivotal supplier analysis that indicates the frequency with which a supplier was pivotal
increased at higher levels of demand. This is consistent with observations in prior years. To
evaluate participant conduct we estimate measures of physical and economic withholding. We
examine withholding patterns relative to the level of demand and the size of each supplier’s
portfolio.
Based on these analyses, we find the overall performance of the ERCOT wholesale market to be
competitive in 2012.
A.
Structural Market Power Indicators
We analyze market structure by using the Residual Demand Index (“RDI”), a statistic that
measures the percentage of load that could not be satisfied without the resources of the largest
supplier. The RDI is used to measure the percentage of load that cannot be served without the
resources of the largest supplier, assuming that the market could call upon all committed and
quick-start capacity owned by other suppliers. 27 When the RDI is greater than zero, the largest
supplier is pivotal (i.e., its resources are needed to satisfy the market demand). When the RDI is
less than zero, no single supplier’s resources are required to serve the load as long as the
resources of its competitors are available.
The RDI is a useful structural indicator of potential market power, although it is important to
recognize its limitations. As a structural indicator, it does not illuminate actual supplier behavior
to indicate whether a supplier may have exercised market power. The RDI also does not indicate
whether it would have been profitable for a pivotal supplier to exercise market power. However,
27
For the purpose of this analysis, “quick-start” includes off-line simple cycle gas turbines that are flagged as online in the current operating plan with a planned generation level of 0 MW that ERCOT has identified as
capable of starting-up and reaching full output after receiving a dispatch instruction from the real-time energy
market.
Page 95
Analysis of Competitive Performance
ERCOT 2012 State of the Market Report
it does identify conditions under which a supplier would have the ability to raise prices
significantly by withholding resources.
Figure 71 shows the RDI relative to load for all hours in 2012. The trend line indicates a strong
positive relationship between load and the RDI. This analysis shown below is done at the QSE
level because the largest suppliers that determine the RDI values own a large majority of the
resources they are offering. It is possible that they also control other capacity through bilateral
arrangements, although we do not know whether this is the case. To the extent that the resources
scheduled by the largest QSEs are not controlled or providing revenue to the QSE, the RDIs will
tend to be slightly overstated.
Figure 71: Residual Demand Index
Figure 72 below summarizes the results of our RDI analysis by displaying the percent of time at
each load level there was a pivotal supplier. At loads greater than 65 GW there was a pivotal
supplier 100 percent of the time. The figure also displays the percent of time each load level
occurs. Combining these values we find that there was a pivotal supplier in approximately
Page 96
ERCOT 2012 State of the Market Report
Analysis of Competitive Performance
12 percent of all hours of 2012. As a comparison, the same system-wide measure for the
Midwest ISO resulted in less than 1 percent of all hours with a pivotal supplier.
Figure 72: Pivotal Supplier Frequency by Load Level
50%
100%
45%
90%
Percent of Hours with Pivotal Supplier
80%
Percent of Time at Load Level
35%
70%
30%
60%
25%
50%
20%
40%
15%
30%
10%
20%
5%
10%
0%
Percent of Hours with Pivotal Supplier
Percent of Hours at Load Levels
40%
0%
<25
25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 60 - 65
Load (GW)
>65
It is important to recognize that inferences regarding market power cannot be made solely from
this data. Bilateral contract obligations can affect a supplier’s potential market power. For
example, a smaller supplier selling energy in the real-time energy market and through short-term
bilateral contracts may have a much greater incentive to exercise market power than a larger
supplier with substantial long-term sales contracts. The RDI measure shown in the previous
figure does not consider the contractual position of the supplier, which can increase a supplier’s
incentive to exercise market power compared to the load-adjusted capacity assumption made in
this analysis.
Page 97
Analysis of Competitive Performance
ERCOT 2012 State of the Market Report
In the next analysis of RDI, we impose ramp rate limitations on the capacity available to meet
load. As shown in Figure 73, the ramp constrained RDI shows the same pattern of becoming
increasingly positive at higher load levels, but is much more likely to be positive as the total
capacity available to the market is smaller than in the previous analysis. We observe that the
ramp rate constrained RDI was usually positive, indicating the presence of a pivotal supplier,
except when load was below 25 GW.
Figure 73: Ramp-Constrained Residual Demand Index
Figure 74 displays the percent of time at each load level there was a pivotal supplier when ramp
rate constraints are considered. At loads greater than approximately 50 GW there is a pivotal
supplier 100 percent of the time. Ramp rate constrained RDI indicates that there was a pivotal
supplier in approximately 80 percent of all hours in 2012. It is important to note that this ramp
rate constraint is being imposed for every dispatch interval, or approximately every 5 minutes.
Page 98
ERCOT 2012 State of the Market Report
Analysis of Competitive Performance
Figure 74: Frequency of Ramp Constrained Pivotal Supplier by Load Level
Percent of Hours at Load Level
45%
100%
Percent of Hours with Pivotal Supplier
Percent of Hours at Load Level
90%
40%
80%
35%
70%
30%
60%
25%
50%
20%
40%
15%
30%
10%
20%
5%
10%
0%
Percent of Hours with Pivotal Supplier
50%
0%
<25
25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 60 - 65
Load (GW)
>65
Voluntary Mitigation Plans
The PUCT approved Voluntary Mitigation Plans (“VMP”) for two market participants – NRG
and GDF SUEZ – during 2012. Action on the request to approve a VMP for a third participant,
Calpine, was pending at the end of the year. Generation owners are motivated to enter into
VMPs because adherence to a plan approved by the commission constitutes an absolute defense
against an allegation of market power abuse through economic withholding with respect to
behaviors addressed by the plan. This increased regulatory certainty afforded to a generation
owner regarding its energy offers in the ERCOT real-time market, must be balanced by
appropriate protections against a potential abuse of market power in violation of PURA
§39.157(a) and PUCT SUBST. R. 25.503(g)(7).
It is our position that VMPs should promote competitive outcomes and prevent abuse of market
power in the ERCOT real-time energy market through economic withholding. The same
restrictions are not required in forward energy markets (e.g., the ERCOT day-ahead market)
because the price in forward energy markets is derived from the real-time energy prices. Because
Page 99
Analysis of Competitive Performance
ERCOT 2012 State of the Market Report
the forward energy markets are voluntary and the market rules do not inhibit arbitrage between
the forward energy markets and the real-time energy market, competitive outcomes in the realtime energy market serve to discipline the potential abuse of market power in the forward energy
markets.
The plan approved for NRG allows the company to offer some of its capacity at prices up to the
system-wide offer cap. Specifically, up to 12 percent of the difference between the high
sustained limit and the low sustained limit for each natural gas-fired unit (5 percent for each
coal/lignite unit) may be offered no higher than the higher of $500 per MWh or 50 times the
natural gas price. Additionally, up to 3 percent of the difference between the high sustained limit
and the low sustained limit for each natural gas-fired unit may be offered no higher than the
system-wide offer cap. The amount of capacity covered by these provisions could be as much as
400 MW.
Allowing offers up to these high levels is intended to accommodate potential legitimate
fluctuations in marginal cost that may exceed the base offer caps, such as operational risks,
short-term fluctuations in fuel costs or availability, or other factors. However, NRG’s VMP
contains a requirement that these offers, if offered in any hour of an operating day, must be
offered in the same price/quantity pair for all hours of the operating day. This provision, along
with the quantity limitations, significantly reduces the potential for tuning these offers in
response to particular market conditions and significantly increases the likelihood that such
offers, if offered, are based on legitimate marginal cost considerations.
Under P.U.C. Subst. R. §25.505(d), market participants controlling less than five percent of the
capacity in ERCOT by definition do not possess ERCOT-wide market power under the PUCT
rules. Hence, these participants can submit very high-priced offers that, per the PUCT rule, will
not be deemed to be an exercise of market power. Although 5 percent of total ERCOT capacity
may seem like a small amount, the potential market impacts of a market participant whose size is
just under the 5 percent threshold choosing to exercise flexibility and offering a significant
portion of their fleet at very high prices could be large.
The figure below shows the amount of surplus capacity available in each hour of every day
during 2011 and 2012. For this analysis, surplus capacity is defined as online generation plus
Page 100
ERCOT 2012 State of the Market Report
Analysis of Competitive Performance
any offline capacity that was available day ahead, plus DC Tie imports (minus exports), minus
responsive reserves provided by generation and regulation up capacity, minus load. Over the
past two years there were 13 hours with no surplus capacity. These correspond to times when
ERCOT was unable to meet load and maintain all operating reserve obligations. Currently the 5
percent “small fish” threshold is roughly 4,000 MW, as indicated by the red line in Figure 75.
There were 450 hours over the past two years with less than 4,000 MW of surplus capacity.
During these times a large “small fish” would be pivotal and able through their offers to increase
the market clearing price, potentially as high as the system-wide offer cap.
Figure 75: Surplus Capacity
To date, the over-mitigation issue discussed in Section I.E, Mitigation at page 20 has meant that
mitigation measures have been applied much more broadly than intended or necessary in the
ERCOT real-time energy market. Market system changes to narrow the scope of mitigation are
scheduled to be implemented in June 2013 to address this issue. Although “small fish” market
participants have always been allowed to offer up to all their capacity at prices up to the system-
Page 101
Analysis of Competitive Performance
ERCOT 2012 State of the Market Report
wide offer cap, the effect on market outcomes of a large “small fish” offering substantial
quantities at high prices will become more noticeable after the scope of mitigation is narrowed.
The approved NRG VMP affords the company offer flexibility for up to approximately
400 MW. 28 As indicated by the green line in Figure 75, the ability for NRG to raise the clearing
price as a result of its offers would have occurred in less than 30 hours over the past two years.
The final key element in a VMP is the timing of termination. The approved VMP for NRG may
be terminated after three business days’ notice. PURA §39.157(a) defines market power abuses
as "practices by persons possessing market power that are unreasonably discriminatory or tend to
unreasonably restrict, impair, or reduce the level of competition..." The exercise of market
power may not rise to the level of an abuse of market power if it does not unreasonably impair
competition, which would involve profitably raising prices significantly above the competitive
level for a significant period of time. Thus, although the offer thresholds provided in the VMP
are designed based on experience to promote competitive market outcomes, the short termination
provision provides additional assurance that any unintended consequences associated with the
potential exercise of market power can be addressed in a timely manner rather than persisting
and rising to the level of an abuse of market power.
B.
Evaluation of Supplier Conduct
The previous subsection presented a structural analysis that supports inferences about potential
market power. In this section we evaluate actual participant conduct to assess whether market
participants have attempted to exercise market power through physical or economic withholding.
First, we examine unit deratings and forced outages to detect physical withholding and then we
evaluate the “output gap” to detect economic withholding.
28
Under the terms of their VMP, NRG may offer a certain portion of their dispatchable capacity from online units at
prices up to $500 per MWH – 5 percent of coal units and 12 percent of gas units. Additionally, NRG may offer up
to 3 percent of their dispatchable capacity from online gas units at prices up to the system-wide offer cap. Any
capacity offered under either of these terms must be offered in the same price/quantity pairs for all hours of the
operating day.
Page 102
ERCOT 2012 State of the Market Report
Analysis of Competitive Performance
In a single-price auction like the real-time energy market, suppliers may attempt to exercise
market power by withholding resources. The purpose of withholding is to cause more expensive
resources to set higher market clearing prices, allowing the supplier to profit on its other sales in
the real-time energy market. Because forward prices will generally be highly correlated with
spot prices, price increases in the real-time energy market can also increase a supplier’s profits in
the bilateral energy market. The strategy is profitable only if the withholding firm’s incremental
profit due to higher price is greater than the lost profit from the foregone sales of its withheld
capacity.
1.
Generation Outages and Deratings
A substantial portion of the installed capability is frequently unavailable due to generator outages
and deratings. For this analysis we start with the unit status information communicated to
ERCOT on a continuous basis. For those units with a status of OUT, meaning they are
unavailable, we then cross check to see if an outage had been scheduled. If there is a
corresponding scheduled outage we consider the unit on planned outage. If not, it is considered
to be a forced outage. We further define derated capacity as the difference between the
summertime maximum capability of a generating resource and its actual capability as
communicated to ERCOT on a continuous basis. It is very common for generating capacity to
be partially derated (e.g., by 5 to 10 percent) because the resource cannot achieve its installed
capability level due to technical or environmental factors (e.g., component equipment failures or
ambient temperature conditions). It is rare for wind generators to produce at their installed
capacity rating due to variations in available wind input. Because such a large portion of derated
capacity is related to wind generation we show it separately. In this subsection, we evaluate
long-term and short-term deratings to inform our evaluation of ERCOT capacity levels.
Figure 76 shows a breakdown of total installed capability for ERCOT on a daily basis during
2012. This analysis includes all in-service and switchable capacity. From the total installed
capacity we subtract away (a) capacity from private networks not available for export to the
ERCOT grid, (b) wind capacity not available due to the lack of wind input, (c) short-term
deratings, (d) short-term planned outages, (e) short-term forced outages, and (e) long-term –
greater than 30 day -- outages and deratings. What remains is the capacity available to serve
load.
Page 103
Analysis of Competitive Performance
ERCOT 2012 State of the Market Report
Outages and deratings of non-wind generators fluctuated between 3 and 18 GW, as shown in
Figure 76, while wind unavailability varied between 1 and 10 GW. Short term planned outages
were largest in March, April and October and small during the summer, which are consistent
with expectations. Short term forced outages also declined during the summer. Short term
deratings peaked during September.
Figure 76: Reductions in Installed Capability
90
Installed Capacity
Private Network capacity not available for export to the ERCOT grid
85
Generation Capacity (GW)
80
75
70
65
60
55
Unavailable Wind
Short Term Planned Outage
Long Term Unavailable
50
Jan
Feb
Mar
Apr
May
Short Term Derating
Short Term Forced Outage
Capacity Available To Serve Load
Jun
Jul
Aug
Sep
Oct
Nov
Dec
The quantity of long term (greater than 30 days) unavailable capacity, peaked in March at nearly
10GW, reduced to 2 GW during the summer months and increased to almost 6GW in October.
This pattern reflects the choice by some owners to mothball certain generators on a seasonal
basis, maintaining the units’ operational status only during the high load summer season when
more costly units have a higher likelihood of operating.
Page 104
ERCOT 2012 State of the Market Report
Analysis of Competitive Performance
The next analysis focuses specifically on short-term planned and forced outages and deratings.
Figure 77 shows the average magnitude of the outages and deratings lasting less than 30 days for
the year and for each month during 2012.
Figure 77: Short-Term Outages and Deratings
10%
Deratings
9%
Planned Outages
Percent of Generating Capacity
8%
Forced Outages
7%
6%
5%
4%
3%
2%
1%
0%
2012
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Figure 77 shows that total short-term deratings and outages were as large as 8 percent of installed
capacity in January, dropping to below 4 percent for the summer. Most of this fluctuation was
due to anticipated planned outages. The amount of capacity unavailable during 2012 averaged
slightly above 5 percent of installed capacity. This is a decrease from 2011, when the amount
was greater than 6 percent. Similar metrics from the zonal market were consistently above
15 percent. The large disparity between values from the zonal and nodal markets is likely due to
combined effects of improved incentives in the nodal market and the lack of unit specific data
available from zonal market systems.
2.
Evaluation of Potential Physical Withholding
Physical withholding occurs when a participant makes resources unavailable for dispatch that are
otherwise physically capable of providing energy and that are economic at prevailing market
Page 105
Analysis of Competitive Performance
ERCOT 2012 State of the Market Report
prices. This can be done either by derating a unit or declaring it as forced out of service.
Because generator deratings and forced outages are unavoidable, the goal of the analysis in this
section is to differentiate justifiable deratings and outages from physical withholding. We test
for physical withholding by examining deratings and outage data to ascertain whether the data
are correlated with conditions under which physical withholding would likely be most profitable.
The RDI results shown in Figure 71 through Figure 74 indicate that the potential for market
power abuse rises at higher load levels as the frequency of positive RDI values increases. Hence,
if physical withholding is a problem in ERCOT, we would expect to see increased deratings and
outages at the highest load levels. Conversely, because competitive prices increase as load
increases, deratings and outages in a market performing competitively will tend to decrease as
load approaches peak levels. Suppliers that lack market power will take actions to maximize the
availability of their resources since their output is generally most profitable in these peak
periods.
Figure 78: Outages and Deratings by Load Level and Participant Size
June to August, 2012
10%
Derating
9%
Planned Outage
Average Percent of Capacity
8%
Forced Outage
7%
6%
5%
4%
3%
2%
1%
<25 GW 25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 60 - 65
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
Large
Small
0%
>65
Figure 78 shows the average relationship of short-term deratings and forced outages as a
percentage of total installed capacity to real-time load level for large and small suppliers.
Page 106
ERCOT 2012 State of the Market Report
Analysis of Competitive Performance
Portfolio size is important in determining whether individual suppliers have incentives to
withhold available resources. Hence, the patterns of outages and deratings of large suppliers can
be usefully evaluated by comparing them to the small suppliers’ patterns.
Long-term deratings are not included in this analysis because they are unlikely to constitute
physical withholding given the cost of such withholding. Wind and private network resources
are also excluded from this analysis because of the high variation in the availability of these
classes of resources. The large supplier category includes the four largest suppliers in ERCOT.
The small supplier category includes the remaining suppliers.
Figure 78 suggests that as demand for electricity increases, all market participants tend to make
more capacity available to the market. For both small and large suppliers, the combined shortterm derating and forced outage rates decreased from 6 to 7 percent at low demand levels to less
than 3 percent at load levels above 65 GW. We observe that at all load levels the percent of
unavailable capacity from large suppliers is less than that from small suppliers.
3.
Evaluation of Potential Economic Withholding
To complement the prior analysis of physical withholding, this subsection evaluates potential
economic withholding by calculating an “output gap”. The output gap is defined as the quantity
of energy that is not being produced by in-service capacity even though the in-service capacity is
economic by a substantial margin given the real-time energy price. A participant can
economically withhold resources, as measured by the output gap, by raising its energy offers so
as not to be dispatched.
Resources are considered for inclusion in the output gap when they are committed and producing
at less than full output. Energy not produced from committed resources is included in the output
gap if the real-time energy price exceeds by at least $50 per MWh that unit’s mitigated offer cap,
which serves as an estimate of the marginal production cost of energy from that resource.
Before presenting the results of the Output Gap analysis, a description of the two-step aspect of
ERCOT’s dispatch software is required. In the first step, the dispatch software calculates output
levels (Base Points) and associated locational marginal prices using the participants’ offer curves
and only considering transmission constraints that have been deemed competitive. These
Page 107
Analysis of Competitive Performance
ERCOT 2012 State of the Market Report
“reference prices” at each generator location are compared with that generator’s mitigated offer
cap, and the higher of the two is used to formulate the offer curve to be used for that generator in
the second step in the dispatch process. The resulting mitigated offer curve is used by the
dispatch software to determine the final output levels for each generator, taking all transmission
constraints into consideration.
If a market participant has sufficient market power, it might raise its offer in such a way to
increase the reference price in the first step. Although in the second step the offer appears to be
mitigated, the market participant has still influenced the market price. This output gap is
measured by the difference between the capacity level on a generator’s original offer curve at the
first step reference price and the capacity level on the generator’s cost curve at the first step
reference price. However, this output gap is only indicative because no output instructions are
sent based on the first step. It is only used to screen out whether a market participant is
withholding in a manner that may influence the reference price.
Figure 79: Incremental Output Gap by Load Level and Participant Size – Step 1
1.0%
Small
Large
Average Percent of Capacity
0.8%
0.6%
0.4%
0.2%
0.0%
<25
Page 108
25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 60 - 65
Load (GW)
>65
ERCOT 2012 State of the Market Report
Analysis of Competitive Performance
From the results of this analysis, shown in Figure 79, we observe only very small amounts of
capacity at only the very highest loads that would be considered part of this output gap. These
small quantities raise no competitive concerns.
Figure 80 shows the ultimate output gap, measured by the difference between a unit’s operating
level and the output level had the unit been competitively offered to the market. In the second
step of the dispatch, the after-mitigation offer curve is used to determine dispatch instructions
and locational prices. As previously illustrated, even though the offer curve is mitigated there is
still the potential for the mitigated offer curve to be increased as a result of a high first step
reference price due to a market participant exerting market power.
Similar to the previous analysis, Figure 80 shows the magnitude of the output gap to be very
small, even at the highest load levels. These small quantities raise no competitive concerns.
Figure 80: Incremental Output Gap by Load Level and Participant Size – Step 2
1.0%
Small
Large
Average Percent of Capacity
0.8%
0.6%
0.4%
0.2%
0.0%
<25
25 - 30 30 - 35 35 - 40 40 - 45 45 - 50 50 - 55 55 - 60 60 - 65
Load (GW)
>65
Page 109